diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12430 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 45.04098729844158,
+  "global_step": 1000000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.940000000000001e-07,
+      "loss": 2.7675,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9940000000000003e-06,
+      "loss": 2.3614,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.994e-06,
+      "loss": 2.1968,
+      "step": 1500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.994e-06,
+      "loss": 2.0825,
+      "step": 2000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.994000000000001e-06,
+      "loss": 2.0089,
+      "step": 2500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9940000000000005e-06,
+      "loss": 1.9457,
+      "step": 3000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.994000000000001e-06,
+      "loss": 1.8901,
+      "step": 3500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.994e-06,
+      "loss": 1.8554,
+      "step": 4000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.994e-06,
+      "loss": 1.8225,
+      "step": 4500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.994000000000001e-06,
+      "loss": 1.7889,
+      "step": 5000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.0994e-05,
+      "loss": 1.7619,
+      "step": 5500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.1994e-05,
+      "loss": 1.7383,
+      "step": 6000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.2994000000000003e-05,
+      "loss": 1.7211,
+      "step": 6500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3992000000000001e-05,
+      "loss": 1.6972,
+      "step": 7000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.4992000000000001e-05,
+      "loss": 1.6789,
+      "step": 7500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.5992000000000002e-05,
+      "loss": 1.6595,
+      "step": 8000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6992000000000003e-05,
+      "loss": 1.6455,
+      "step": 8500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7992e-05,
+      "loss": 1.6282,
+      "step": 9000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8992e-05,
+      "loss": 1.6182,
+      "step": 9500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9992e-05,
+      "loss": 1.6009,
+      "step": 10000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.999626258363976e-05,
+      "loss": 1.5884,
+      "step": 10500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9992510097052265e-05,
+      "loss": 1.5719,
+      "step": 11000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9988742540237507e-05,
+      "loss": 1.5618,
+      "step": 11500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9984974983422753e-05,
+      "loss": 1.551,
+      "step": 12000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9981207426607995e-05,
+      "loss": 1.5427,
+      "step": 12500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9977439869793237e-05,
+      "loss": 1.5336,
+      "step": 13000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.997367231297848e-05,
+      "loss": 1.5236,
+      "step": 13500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9969904756163725e-05,
+      "loss": 1.5199,
+      "step": 14000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9966137199348967e-05,
+      "loss": 1.5082,
+      "step": 14500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.996237717764784e-05,
+      "loss": 1.5066,
+      "step": 15000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9958609620833084e-05,
+      "loss": 1.4972,
+      "step": 15500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9954842064018326e-05,
+      "loss": 1.49,
+      "step": 16000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.995107450720357e-05,
+      "loss": 1.4823,
+      "step": 16500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9947306950388813e-05,
+      "loss": 1.4791,
+      "step": 17000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9943539393574055e-05,
+      "loss": 1.4753,
+      "step": 17500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.993977937187293e-05,
+      "loss": 1.468,
+      "step": 18000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9936011815058172e-05,
+      "loss": 1.4591,
+      "step": 18500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.9932244258243418e-05,
+      "loss": 1.4554,
+      "step": 19000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.992848423654229e-05,
+      "loss": 1.4535,
+      "step": 19500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.992471667972753e-05,
+      "loss": 1.4466,
+      "step": 20000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9920949122912776e-05,
+      "loss": 1.444,
+      "step": 20500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.991718156609802e-05,
+      "loss": 1.44,
+      "step": 21000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.991341400928326e-05,
+      "loss": 1.4354,
+      "step": 21500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9909646452468506e-05,
+      "loss": 1.4303,
+      "step": 22000
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7359925975935765,
+      "eval_loss": 1.2059218883514404,
+      "eval_runtime": 57.7689,
+      "eval_samples_per_second": 846.65,
+      "eval_steps_per_second": 13.242,
+      "step": 22202
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9905878895653748e-05,
+      "loss": 1.427,
+      "step": 22500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9902111338838994e-05,
+      "loss": 1.4173,
+      "step": 23000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9898343782024236e-05,
+      "loss": 1.4192,
+      "step": 23500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9894576225209478e-05,
+      "loss": 1.4152,
+      "step": 24000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.989081620350835e-05,
+      "loss": 1.4104,
+      "step": 24500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.9887048646693595e-05,
+      "loss": 1.4078,
+      "step": 25000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9883281089878837e-05,
+      "loss": 1.4046,
+      "step": 25500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9879513533064082e-05,
+      "loss": 1.402,
+      "step": 26000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9875753511362954e-05,
+      "loss": 1.3987,
+      "step": 26500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9871985954548196e-05,
+      "loss": 1.3946,
+      "step": 27000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.986822593284707e-05,
+      "loss": 1.3921,
+      "step": 27500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.9864458376032313e-05,
+      "loss": 1.3915,
+      "step": 28000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9860690819217555e-05,
+      "loss": 1.3875,
+      "step": 28500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.98569232624028e-05,
+      "loss": 1.3862,
+      "step": 29000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9853155705588042e-05,
+      "loss": 1.3799,
+      "step": 29500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9849388148773284e-05,
+      "loss": 1.3801,
+      "step": 30000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9845620591958526e-05,
+      "loss": 1.3773,
+      "step": 30500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.984185303514377e-05,
+      "loss": 1.3751,
+      "step": 31000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9838085478329014e-05,
+      "loss": 1.3749,
+      "step": 31500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.983431792151426e-05,
+      "loss": 1.3733,
+      "step": 32000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.98305503646995e-05,
+      "loss": 1.3651,
+      "step": 32500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9826782807884743e-05,
+      "loss": 1.364,
+      "step": 33000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9823022786183618e-05,
+      "loss": 1.3656,
+      "step": 33500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.981925522936886e-05,
+      "loss": 1.3643,
+      "step": 34000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9815487672554102e-05,
+      "loss": 1.3589,
+      "step": 34500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9811720115739348e-05,
+      "loss": 1.355,
+      "step": 35000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.980796009403822e-05,
+      "loss": 1.357,
+      "step": 35500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.980420007233709e-05,
+      "loss": 1.3531,
+      "step": 36000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.9800432515522336e-05,
+      "loss": 1.3501,
+      "step": 36500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9796664958707578e-05,
+      "loss": 1.3494,
+      "step": 37000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.9792897401892824e-05,
+      "loss": 1.3482,
+      "step": 37500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.9789129845078066e-05,
+      "loss": 1.3461,
+      "step": 38000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.9785369823376937e-05,
+      "loss": 1.3439,
+      "step": 38500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.9781602266562183e-05,
+      "loss": 1.3399,
+      "step": 39000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.9777834709747425e-05,
+      "loss": 1.3422,
+      "step": 39500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9774067152932667e-05,
+      "loss": 1.3383,
+      "step": 40000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.977029959611791e-05,
+      "loss": 1.3337,
+      "step": 40500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9766532039303154e-05,
+      "loss": 1.3335,
+      "step": 41000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9762764482488396e-05,
+      "loss": 1.3337,
+      "step": 41500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.9758996925673642e-05,
+      "loss": 1.3306,
+      "step": 42000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.9755236903972513e-05,
+      "loss": 1.3294,
+      "step": 42500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.9751476882271385e-05,
+      "loss": 1.3258,
+      "step": 43000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.974770932545663e-05,
+      "loss": 1.3252,
+      "step": 43500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.9743941768641872e-05,
+      "loss": 1.3219,
+      "step": 44000
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.7503807978708374,
+      "eval_loss": 1.1156708002090454,
+      "eval_runtime": 57.6366,
+      "eval_samples_per_second": 848.592,
+      "eval_steps_per_second": 13.273,
+      "step": 44404
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.9740174211827118e-05,
+      "loss": 1.3249,
+      "step": 44500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.973640665501236e-05,
+      "loss": 1.3209,
+      "step": 45000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.9732639098197605e-05,
+      "loss": 1.3188,
+      "step": 45500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.9728879076496473e-05,
+      "loss": 1.3184,
+      "step": 46000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.9725119054795348e-05,
+      "loss": 1.3158,
+      "step": 46500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.9721351497980593e-05,
+      "loss": 1.3149,
+      "step": 47000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.9717583941165835e-05,
+      "loss": 1.3127,
+      "step": 47500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.9713816384351078e-05,
+      "loss": 1.311,
+      "step": 48000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.971004882753632e-05,
+      "loss": 1.3109,
+      "step": 48500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.9706281270721565e-05,
+      "loss": 1.306,
+      "step": 49000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.9702513713906807e-05,
+      "loss": 1.3064,
+      "step": 49500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.969874615709205e-05,
+      "loss": 1.3053,
+      "step": 50000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.9694986135390924e-05,
+      "loss": 1.3029,
+      "step": 50500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.9691218578576166e-05,
+      "loss": 1.3025,
+      "step": 51000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.968745102176141e-05,
+      "loss": 1.3024,
+      "step": 51500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.9683691000060283e-05,
+      "loss": 1.3003,
+      "step": 52000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.9679923443245525e-05,
+      "loss": 1.2994,
+      "step": 52500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.967615588643077e-05,
+      "loss": 1.298,
+      "step": 53000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9672388329616013e-05,
+      "loss": 1.2953,
+      "step": 53500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9668620772801255e-05,
+      "loss": 1.2948,
+      "step": 54000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.96648532159865e-05,
+      "loss": 1.2949,
+      "step": 54500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.9661085659171742e-05,
+      "loss": 1.293,
+      "step": 55000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.9657318102356988e-05,
+      "loss": 1.2899,
+      "step": 55500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.965355054554223e-05,
+      "loss": 1.2942,
+      "step": 56000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.9649782988727472e-05,
+      "loss": 1.2874,
+      "step": 56500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.9646015431912714e-05,
+      "loss": 1.2919,
+      "step": 57000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.9642247875097956e-05,
+      "loss": 1.289,
+      "step": 57500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.96384803182832e-05,
+      "loss": 1.2854,
+      "step": 58000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.9634720296582076e-05,
+      "loss": 1.2861,
+      "step": 58500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.9630952739767318e-05,
+      "loss": 1.2867,
+      "step": 59000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.962718518295256e-05,
+      "loss": 1.2847,
+      "step": 59500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.9623417626137802e-05,
+      "loss": 1.2821,
+      "step": 60000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.9619650069323048e-05,
+      "loss": 1.2832,
+      "step": 60500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.961588251250829e-05,
+      "loss": 1.2778,
+      "step": 61000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.9612114955693532e-05,
+      "loss": 1.2769,
+      "step": 61500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.9608354933992407e-05,
+      "loss": 1.2794,
+      "step": 62000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.960458737717765e-05,
+      "loss": 1.2767,
+      "step": 62500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.960082735547652e-05,
+      "loss": 1.2747,
+      "step": 63000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.9597059798661766e-05,
+      "loss": 1.2755,
+      "step": 63500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9593292241847008e-05,
+      "loss": 1.2751,
+      "step": 64000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.9589524685032253e-05,
+      "loss": 1.2738,
+      "step": 64500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.9585757128217495e-05,
+      "loss": 1.2754,
+      "step": 65000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.9581989571402737e-05,
+      "loss": 1.2698,
+      "step": 65500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.9578222014587983e-05,
+      "loss": 1.2708,
+      "step": 66000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9574454457773225e-05,
+      "loss": 1.2688,
+      "step": 66500
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.757473410577184,
+      "eval_loss": 1.0706008672714233,
+      "eval_runtime": 58.6548,
+      "eval_samples_per_second": 833.861,
+      "eval_steps_per_second": 13.042,
+      "step": 66606
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.9570694436072096e-05,
+      "loss": 1.2663,
+      "step": 67000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.956693441437097e-05,
+      "loss": 1.2653,
+      "step": 67500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.9563166857556213e-05,
+      "loss": 1.2651,
+      "step": 68000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.955939930074146e-05,
+      "loss": 1.2652,
+      "step": 68500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.95556317439267e-05,
+      "loss": 1.2667,
+      "step": 69000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.9551864187111943e-05,
+      "loss": 1.2643,
+      "step": 69500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.9548096630297185e-05,
+      "loss": 1.2625,
+      "step": 70000
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.954433660859606e-05,
+      "loss": 1.2584,
+      "step": 70500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.9540569051781302e-05,
+      "loss": 1.2595,
+      "step": 71000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9536801494966547e-05,
+      "loss": 1.2615,
+      "step": 71500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.953303393815179e-05,
+      "loss": 1.262,
+      "step": 72000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.9529266381337035e-05,
+      "loss": 1.2573,
+      "step": 72500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.9525498824522277e-05,
+      "loss": 1.258,
+      "step": 73000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.952173126770752e-05,
+      "loss": 1.2579,
+      "step": 73500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.951796371089276e-05,
+      "loss": 1.2535,
+      "step": 74000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.9514203689191636e-05,
+      "loss": 1.2555,
+      "step": 74500
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.9510436132376878e-05,
+      "loss": 1.2563,
+      "step": 75000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.9506668575562123e-05,
+      "loss": 1.251,
+      "step": 75500
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.9502901018747365e-05,
+      "loss": 1.2546,
+      "step": 76000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.9499133461932607e-05,
+      "loss": 1.2552,
+      "step": 76500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.9495365905117853e-05,
+      "loss": 1.2512,
+      "step": 77000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.9491598348303095e-05,
+      "loss": 1.2532,
+      "step": 77500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.9487830791488337e-05,
+      "loss": 1.2481,
+      "step": 78000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.948406323467358e-05,
+      "loss": 1.2502,
+      "step": 78500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.9480295677858824e-05,
+      "loss": 1.248,
+      "step": 79000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9476528121044067e-05,
+      "loss": 1.2496,
+      "step": 79500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.947276809934294e-05,
+      "loss": 1.2481,
+      "step": 80000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.9469000542528183e-05,
+      "loss": 1.2458,
+      "step": 80500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.9465232985713425e-05,
+      "loss": 1.2452,
+      "step": 81000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.9461472964012297e-05,
+      "loss": 1.2434,
+      "step": 81500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.9457705407197542e-05,
+      "loss": 1.2454,
+      "step": 82000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.9453937850382784e-05,
+      "loss": 1.2433,
+      "step": 82500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.945017029356803e-05,
+      "loss": 1.2393,
+      "step": 83000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.9446402736753272e-05,
+      "loss": 1.2408,
+      "step": 83500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.9442642715052143e-05,
+      "loss": 1.2389,
+      "step": 84000
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.9438875158237385e-05,
+      "loss": 1.2374,
+      "step": 84500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.943510760142263e-05,
+      "loss": 1.2369,
+      "step": 85000
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.9431340044607873e-05,
+      "loss": 1.2401,
+      "step": 85500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.942757248779312e-05,
+      "loss": 1.2351,
+      "step": 86000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.942380493097836e-05,
+      "loss": 1.2385,
+      "step": 86500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.9420037374163606e-05,
+      "loss": 1.2335,
+      "step": 87000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.9416269817348848e-05,
+      "loss": 1.2356,
+      "step": 87500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.941250979564772e-05,
+      "loss": 1.2356,
+      "step": 88000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.8414808080808082e-05,
+      "loss": 1.2313,
+      "step": 88500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.8404707070707074e-05,
+      "loss": 1.2344,
+      "step": 89000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.8394606060606063e-05,
+      "loss": 1.2331,
+      "step": 89500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.838450505050505e-05,
+      "loss": 1.233,
+      "step": 90000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.8374424242424245e-05,
+      "loss": 1.2321,
+      "step": 90500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.8364323232323234e-05,
+      "loss": 1.2329,
+      "step": 91000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.8354222222222226e-05,
+      "loss": 1.2277,
+      "step": 91500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.8344141414141416e-05,
+      "loss": 1.2293,
+      "step": 92000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.8334040404040405e-05,
+      "loss": 1.2281,
+      "step": 92500
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8323939393939397e-05,
+      "loss": 1.2254,
+      "step": 93000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8313838383838385e-05,
+      "loss": 1.2264,
+      "step": 93500
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.8303737373737377e-05,
+      "loss": 1.2263,
+      "step": 94000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.8293636363636366e-05,
+      "loss": 1.2254,
+      "step": 94500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.8283535353535355e-05,
+      "loss": 1.2253,
+      "step": 95000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.8273434343434343e-05,
+      "loss": 1.222,
+      "step": 95500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.8263333333333335e-05,
+      "loss": 1.224,
+      "step": 96000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.8253252525252526e-05,
+      "loss": 1.2237,
+      "step": 96500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.8243151515151514e-05,
+      "loss": 1.2237,
+      "step": 97000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.8233050505050506e-05,
+      "loss": 1.2243,
+      "step": 97500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.8222949494949495e-05,
+      "loss": 1.2253,
+      "step": 98000
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.8212848484848487e-05,
+      "loss": 1.2227,
+      "step": 98500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.8202747474747476e-05,
+      "loss": 1.221,
+      "step": 99000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.8192666666666666e-05,
+      "loss": 1.2198,
+      "step": 99500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.818258585858586e-05,
+      "loss": 1.2197,
+      "step": 100000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.8172484848484852e-05,
+      "loss": 1.2183,
+      "step": 100500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.816238383838384e-05,
+      "loss": 1.2181,
+      "step": 101000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.815228282828283e-05,
+      "loss": 1.215,
+      "step": 101500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.8142181818181818e-05,
+      "loss": 1.2193,
+      "step": 102000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.813208080808081e-05,
+      "loss": 1.2153,
+      "step": 102500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.81219797979798e-05,
+      "loss": 1.2164,
+      "step": 103000
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.811187878787879e-05,
+      "loss": 1.2172,
+      "step": 103500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.810179797979798e-05,
+      "loss": 1.2134,
+      "step": 104000
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.809169696969697e-05,
+      "loss": 1.2139,
+      "step": 104500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.808159595959596e-05,
+      "loss": 1.2124,
+      "step": 105000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.807149494949495e-05,
+      "loss": 1.2138,
+      "step": 105500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.8061393939393942e-05,
+      "loss": 1.2112,
+      "step": 106000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.8051333333333335e-05,
+      "loss": 1.21,
+      "step": 106500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.8041232323232327e-05,
+      "loss": 1.2114,
+      "step": 107000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.8031131313131315e-05,
+      "loss": 1.2103,
+      "step": 107500
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.8021030303030304e-05,
+      "loss": 1.2122,
+      "step": 108000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.8010929292929292e-05,
+      "loss": 1.2123,
+      "step": 108500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.8000828282828284e-05,
+      "loss": 1.2125,
+      "step": 109000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.7990727272727276e-05,
+      "loss": 1.2097,
+      "step": 109500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.7980626262626265e-05,
+      "loss": 1.2102,
+      "step": 110000
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.7665563896270091,
+      "eval_loss": 1.016308069229126,
+      "eval_runtime": 54.3803,
+      "eval_samples_per_second": 899.407,
+      "eval_steps_per_second": 14.068,
+      "step": 110202
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.7970545454545455e-05,
+      "loss": 1.2068,
+      "step": 110500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.796046464646465e-05,
+      "loss": 1.2054,
+      "step": 111000
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7950363636363638e-05,
+      "loss": 1.2051,
+      "step": 111500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.794028282828283e-05,
+      "loss": 1.207,
+      "step": 112000
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.793018181818182e-05,
+      "loss": 1.2037,
+      "step": 112500
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.792008080808081e-05,
+      "loss": 1.2075,
+      "step": 113000
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.79099797979798e-05,
+      "loss": 1.204,
+      "step": 113500
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.789987878787879e-05,
+      "loss": 1.2064,
+      "step": 114000
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.7889777777777778e-05,
+      "loss": 1.2043,
+      "step": 114500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.787967676767677e-05,
+      "loss": 1.2036,
+      "step": 115000
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.786957575757576e-05,
+      "loss": 1.1989,
+      "step": 115500
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.785947474747475e-05,
+      "loss": 1.2022,
+      "step": 116000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.784939393939394e-05,
+      "loss": 1.2029,
+      "step": 116500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.783929292929293e-05,
+      "loss": 1.2033,
+      "step": 117000
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.7829191919191922e-05,
+      "loss": 1.1998,
+      "step": 117500
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.781909090909091e-05,
+      "loss": 1.201,
+      "step": 118000
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.7808989898989902e-05,
+      "loss": 1.2011,
+      "step": 118500
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.7798909090909093e-05,
+      "loss": 1.1993,
+      "step": 119000
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.778880808080808e-05,
+      "loss": 1.1998,
+      "step": 119500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.7778707070707074e-05,
+      "loss": 1.1997,
+      "step": 120000
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.7768606060606062e-05,
+      "loss": 1.1998,
+      "step": 120500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.7758505050505054e-05,
+      "loss": 1.1997,
+      "step": 121000
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.7748404040404043e-05,
+      "loss": 1.2008,
+      "step": 121500
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.773830303030303e-05,
+      "loss": 1.198,
+      "step": 122000
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7728222222222225e-05,
+      "loss": 1.1941,
+      "step": 122500
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.7718121212121214e-05,
+      "loss": 1.1956,
+      "step": 123000
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.7708020202020203e-05,
+      "loss": 1.1972,
+      "step": 123500
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.7697919191919194e-05,
+      "loss": 1.1988,
+      "step": 124000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.7687818181818183e-05,
+      "loss": 1.195,
+      "step": 124500
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.767771717171717e-05,
+      "loss": 1.1973,
+      "step": 125000
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.7667616161616164e-05,
+      "loss": 1.1946,
+      "step": 125500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7657515151515152e-05,
+      "loss": 1.1933,
+      "step": 126000
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.7647414141414144e-05,
+      "loss": 1.1961,
+      "step": 126500
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.7637313131313133e-05,
+      "loss": 1.1911,
+      "step": 127000
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.762725252525253e-05,
+      "loss": 1.1931,
+      "step": 127500
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.7617151515151517e-05,
+      "loss": 1.1961,
+      "step": 128000
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.7607050505050506e-05,
+      "loss": 1.1926,
+      "step": 128500
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.7596949494949494e-05,
+      "loss": 1.1915,
+      "step": 129000
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7586848484848486e-05,
+      "loss": 1.1916,
+      "step": 129500
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.7576747474747475e-05,
+      "loss": 1.1909,
+      "step": 130000
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7566646464646467e-05,
+      "loss": 1.1882,
+      "step": 130500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.7556545454545456e-05,
+      "loss": 1.1888,
+      "step": 131000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.7546444444444448e-05,
+      "loss": 1.188,
+      "step": 131500
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.7536343434343436e-05,
+      "loss": 1.1878,
+      "step": 132000
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.7690029733422831,
+      "eval_loss": 1.0008476972579956,
+      "eval_runtime": 54.4591,
+      "eval_samples_per_second": 898.104,
+      "eval_steps_per_second": 14.047,
+      "step": 132404
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.7526262626262627e-05,
+      "loss": 1.1892,
+      "step": 132500
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.751616161616162e-05,
+      "loss": 1.1871,
+      "step": 133000
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.7506060606060607e-05,
+      "loss": 1.1854,
+      "step": 133500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.7495959595959596e-05,
+      "loss": 1.1847,
+      "step": 134000
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.748587878787879e-05,
+      "loss": 1.1876,
+      "step": 134500
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.747577777777778e-05,
+      "loss": 1.186,
+      "step": 135000
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.746567676767677e-05,
+      "loss": 1.1872,
+      "step": 135500
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 1.745557575757576e-05,
+      "loss": 1.1851,
+      "step": 136000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 1.7445474747474748e-05,
+      "loss": 1.1856,
+      "step": 136500
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 1.743537373737374e-05,
+      "loss": 1.1838,
+      "step": 137000
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 1.7425272727272728e-05,
+      "loss": 1.1846,
+      "step": 137500
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 1.7415191919191922e-05,
+      "loss": 1.1841,
+      "step": 138000
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 1.740509090909091e-05,
+      "loss": 1.1857,
+      "step": 138500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 1.73949898989899e-05,
+      "loss": 1.1836,
+      "step": 139000
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 1.738488888888889e-05,
+      "loss": 1.1822,
+      "step": 139500
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 1.737478787878788e-05,
+      "loss": 1.183,
+      "step": 140000
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 1.7364686868686872e-05,
+      "loss": 1.1836,
+      "step": 140500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 1.735458585858586e-05,
+      "loss": 1.181,
+      "step": 141000
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 1.734448484848485e-05,
+      "loss": 1.182,
+      "step": 141500
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.7334383838383838e-05,
+      "loss": 1.1799,
+      "step": 142000
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.732428282828283e-05,
+      "loss": 1.1814,
+      "step": 142500
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.7314181818181818e-05,
+      "loss": 1.1797,
+      "step": 143000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.7304101010101012e-05,
+      "loss": 1.1791,
+      "step": 143500
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 1.7294e-05,
+      "loss": 1.1825,
+      "step": 144000
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.728389898989899e-05,
+      "loss": 1.1797,
+      "step": 144500
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 1.727379797979798e-05,
+      "loss": 1.1812,
+      "step": 145000
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.7263717171717175e-05,
+      "loss": 1.1801,
+      "step": 145500
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.7253616161616164e-05,
+      "loss": 1.1774,
+      "step": 146000
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 1.7243535353535354e-05,
+      "loss": 1.1825,
+      "step": 146500
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 1.7233434343434346e-05,
+      "loss": 1.1793,
+      "step": 147000
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 1.7223333333333335e-05,
+      "loss": 1.1783,
+      "step": 147500
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.7213232323232324e-05,
+      "loss": 1.1758,
+      "step": 148000
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 1.7203131313131312e-05,
+      "loss": 1.177,
+      "step": 148500
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 1.7193030303030304e-05,
+      "loss": 1.1772,
+      "step": 149000
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 1.7182929292929293e-05,
+      "loss": 1.1758,
+      "step": 149500
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 1.7172828282828285e-05,
+      "loss": 1.1777,
+      "step": 150000
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 1.7162727272727273e-05,
+      "loss": 1.1728,
+      "step": 150500
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 1.7152626262626265e-05,
+      "loss": 1.1748,
+      "step": 151000
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 1.7142525252525254e-05,
+      "loss": 1.177,
+      "step": 151500
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 1.7132444444444444e-05,
+      "loss": 1.1756,
+      "step": 152000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 1.7122343434343436e-05,
+      "loss": 1.1741,
+      "step": 152500
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 1.7112242424242425e-05,
+      "loss": 1.1767,
+      "step": 153000
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 1.7102141414141417e-05,
+      "loss": 1.1743,
+      "step": 153500
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 1.7092060606060608e-05,
+      "loss": 1.1753,
+      "step": 154000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.7081979797979798e-05,
+      "loss": 1.1723,
+      "step": 154500
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.7708941094427499,
+      "eval_loss": 0.9895298480987549,
+      "eval_runtime": 54.3467,
+      "eval_samples_per_second": 899.963,
+      "eval_steps_per_second": 14.076,
+      "step": 154606
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.7071878787878787e-05,
+      "loss": 1.1718,
+      "step": 155000
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.706177777777778e-05,
+      "loss": 1.1697,
+      "step": 155500
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 1.7051676767676767e-05,
+      "loss": 1.1721,
+      "step": 156000
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.704157575757576e-05,
+      "loss": 1.1704,
+      "step": 156500
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.703147474747475e-05,
+      "loss": 1.1735,
+      "step": 157000
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.702137373737374e-05,
+      "loss": 1.1708,
+      "step": 157500
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.701127272727273e-05,
+      "loss": 1.1702,
+      "step": 158000
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.7001171717171717e-05,
+      "loss": 1.1683,
+      "step": 158500
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.699109090909091e-05,
+      "loss": 1.1669,
+      "step": 159000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.6980989898989903e-05,
+      "loss": 1.1706,
+      "step": 159500
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.697088888888889e-05,
+      "loss": 1.1718,
+      "step": 160000
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.696078787878788e-05,
+      "loss": 1.1714,
+      "step": 160500
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 1.695068686868687e-05,
+      "loss": 1.1675,
+      "step": 161000
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.6940606060606063e-05,
+      "loss": 1.1687,
+      "step": 161500
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.693050505050505e-05,
+      "loss": 1.1684,
+      "step": 162000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.692040404040404e-05,
+      "loss": 1.1675,
+      "step": 162500
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.6910303030303032e-05,
+      "loss": 1.1685,
+      "step": 163000
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.6900222222222226e-05,
+      "loss": 1.1665,
+      "step": 163500
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.6890121212121214e-05,
+      "loss": 1.1684,
+      "step": 164000
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.6880020202020203e-05,
+      "loss": 1.165,
+      "step": 164500
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.686991919191919e-05,
+      "loss": 1.1663,
+      "step": 165000
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.6859818181818184e-05,
+      "loss": 1.1677,
+      "step": 165500
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.6849737373737377e-05,
+      "loss": 1.1657,
+      "step": 166000
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.6839636363636366e-05,
+      "loss": 1.1657,
+      "step": 166500
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.6829535353535355e-05,
+      "loss": 1.165,
+      "step": 167000
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.6819434343434343e-05,
+      "loss": 1.1621,
+      "step": 167500
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.6809353535353537e-05,
+      "loss": 1.1635,
+      "step": 168000
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.6799272727272728e-05,
+      "loss": 1.1642,
+      "step": 168500
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.678917171717172e-05,
+      "loss": 1.164,
+      "step": 169000
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.677907070707071e-05,
+      "loss": 1.1631,
+      "step": 169500
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.67689696969697e-05,
+      "loss": 1.1642,
+      "step": 170000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.675886868686869e-05,
+      "loss": 1.1653,
+      "step": 170500
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.6748767676767678e-05,
+      "loss": 1.1625,
+      "step": 171000
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.6738666666666666e-05,
+      "loss": 1.1629,
+      "step": 171500
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.6728565656565658e-05,
+      "loss": 1.1616,
+      "step": 172000
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.6718484848484852e-05,
+      "loss": 1.1627,
+      "step": 172500
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.670838383838384e-05,
+      "loss": 1.16,
+      "step": 173000
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.669828282828283e-05,
+      "loss": 1.1643,
+      "step": 173500
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.6688181818181818e-05,
+      "loss": 1.1611,
+      "step": 174000
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.667808080808081e-05,
+      "loss": 1.1585,
+      "step": 174500
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.66679797979798e-05,
+      "loss": 1.1572,
+      "step": 175000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.665787878787879e-05,
+      "loss": 1.1591,
+      "step": 175500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.664777777777778e-05,
+      "loss": 1.1609,
+      "step": 176000
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.6637676767676768e-05,
+      "loss": 1.1614,
+      "step": 176500
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.7727673009951773,
+      "eval_loss": 0.9784103631973267,
+      "eval_runtime": 54.0506,
+      "eval_samples_per_second": 904.892,
+      "eval_steps_per_second": 14.153,
+      "step": 176808
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.662759595959596e-05,
+      "loss": 1.159,
+      "step": 177000
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.661749494949495e-05,
+      "loss": 1.1585,
+      "step": 177500
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.6607393939393942e-05,
+      "loss": 1.1582,
+      "step": 178000
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.659729292929293e-05,
+      "loss": 1.156,
+      "step": 178500
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.658719191919192e-05,
+      "loss": 1.1581,
+      "step": 179000
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.6577111111111113e-05,
+      "loss": 1.1577,
+      "step": 179500
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.6567010101010102e-05,
+      "loss": 1.1575,
+      "step": 180000
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.6556909090909094e-05,
+      "loss": 1.159,
+      "step": 180500
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.6546808080808082e-05,
+      "loss": 1.1583,
+      "step": 181000
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.653670707070707e-05,
+      "loss": 1.1555,
+      "step": 181500
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.6526606060606063e-05,
+      "loss": 1.1578,
+      "step": 182000
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.651650505050505e-05,
+      "loss": 1.1574,
+      "step": 182500
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.6506404040404043e-05,
+      "loss": 1.1559,
+      "step": 183000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.6496323232323234e-05,
+      "loss": 1.1557,
+      "step": 183500
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.6486222222222223e-05,
+      "loss": 1.1554,
+      "step": 184000
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.6476121212121215e-05,
+      "loss": 1.1541,
+      "step": 184500
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.6466020202020203e-05,
+      "loss": 1.1559,
+      "step": 185000
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.6455919191919195e-05,
+      "loss": 1.1515,
+      "step": 185500
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.6445838383838386e-05,
+      "loss": 1.156,
+      "step": 186000
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.6435737373737374e-05,
+      "loss": 1.1572,
+      "step": 186500
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.6425636363636366e-05,
+      "loss": 1.1522,
+      "step": 187000
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.6415535353535355e-05,
+      "loss": 1.1526,
+      "step": 187500
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.6405454545454545e-05,
+      "loss": 1.1532,
+      "step": 188000
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.6395353535353537e-05,
+      "loss": 1.1548,
+      "step": 188500
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.6385252525252526e-05,
+      "loss": 1.15,
+      "step": 189000
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.637517171717172e-05,
+      "loss": 1.1519,
+      "step": 189500
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.636507070707071e-05,
+      "loss": 1.1527,
+      "step": 190000
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.6354969696969697e-05,
+      "loss": 1.153,
+      "step": 190500
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.634486868686869e-05,
+      "loss": 1.1527,
+      "step": 191000
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.633478787878788e-05,
+      "loss": 1.1512,
+      "step": 191500
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.6324686868686868e-05,
+      "loss": 1.1509,
+      "step": 192000
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.6314606060606062e-05,
+      "loss": 1.1494,
+      "step": 192500
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.6304505050505054e-05,
+      "loss": 1.149,
+      "step": 193000
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.6294404040404043e-05,
+      "loss": 1.1518,
+      "step": 193500
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.628430303030303e-05,
+      "loss": 1.1508,
+      "step": 194000
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.627420202020202e-05,
+      "loss": 1.1501,
+      "step": 194500
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.6264101010101012e-05,
+      "loss": 1.1487,
+      "step": 195000
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.6254e-05,
+      "loss": 1.1504,
+      "step": 195500
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.6243898989898993e-05,
+      "loss": 1.1489,
+      "step": 196000
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.623379797979798e-05,
+      "loss": 1.1489,
+      "step": 196500
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.6223717171717172e-05,
+      "loss": 1.1484,
+      "step": 197000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.6213616161616164e-05,
+      "loss": 1.1504,
+      "step": 197500
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.6203515151515152e-05,
+      "loss": 1.1514,
+      "step": 198000
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.6193414141414144e-05,
+      "loss": 1.1506,
+      "step": 198500
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.6183333333333335e-05,
+      "loss": 1.1489,
+      "step": 199000
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.7746199929363001,
+      "eval_loss": 0.9665696024894714,
+      "eval_runtime": 54.5357,
+      "eval_samples_per_second": 896.843,
+      "eval_steps_per_second": 14.027,
+      "step": 199010
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.6173232323232323e-05,
+      "loss": 1.1463,
+      "step": 199500
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.6163131313131315e-05,
+      "loss": 1.1455,
+      "step": 200000
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.6153030303030304e-05,
+      "loss": 1.1473,
+      "step": 200500
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.6142929292929296e-05,
+      "loss": 1.1477,
+      "step": 201000
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.6132828282828285e-05,
+      "loss": 1.1469,
+      "step": 201500
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.6122727272727273e-05,
+      "loss": 1.1457,
+      "step": 202000
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.6112626262626262e-05,
+      "loss": 1.1437,
+      "step": 202500
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.6102525252525254e-05,
+      "loss": 1.1497,
+      "step": 203000
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.6092424242424242e-05,
+      "loss": 1.1449,
+      "step": 203500
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.6082323232323234e-05,
+      "loss": 1.1476,
+      "step": 204000
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 1.6072222222222226e-05,
+      "loss": 1.1465,
+      "step": 204500
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 1.6062141414141413e-05,
+      "loss": 1.1422,
+      "step": 205000
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 1.6052060606060607e-05,
+      "loss": 1.1444,
+      "step": 205500
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 1.6041959595959596e-05,
+      "loss": 1.1453,
+      "step": 206000
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 1.6031858585858585e-05,
+      "loss": 1.1454,
+      "step": 206500
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 1.6021757575757577e-05,
+      "loss": 1.1429,
+      "step": 207000
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 1.6011656565656565e-05,
+      "loss": 1.1434,
+      "step": 207500
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 1.6001555555555557e-05,
+      "loss": 1.1478,
+      "step": 208000
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.599145454545455e-05,
+      "loss": 1.1453,
+      "step": 208500
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.598139393939394e-05,
+      "loss": 1.1403,
+      "step": 209000
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.597129292929293e-05,
+      "loss": 1.1434,
+      "step": 209500
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.5961191919191922e-05,
+      "loss": 1.1415,
+      "step": 210000
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.595109090909091e-05,
+      "loss": 1.1459,
+      "step": 210500
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.59409898989899e-05,
+      "loss": 1.1421,
+      "step": 211000
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.593088888888889e-05,
+      "loss": 1.1401,
+      "step": 211500
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.5920808080808082e-05,
+      "loss": 1.1415,
+      "step": 212000
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 1.591070707070707e-05,
+      "loss": 1.1389,
+      "step": 212500
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 1.5900606060606062e-05,
+      "loss": 1.1401,
+      "step": 213000
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.589050505050505e-05,
+      "loss": 1.1413,
+      "step": 213500
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.5880404040404043e-05,
+      "loss": 1.14,
+      "step": 214000
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.587030303030303e-05,
+      "loss": 1.1419,
+      "step": 214500
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.5860202020202024e-05,
+      "loss": 1.1369,
+      "step": 215000
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.5850101010101012e-05,
+      "loss": 1.1402,
+      "step": 215500
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.584e-05,
+      "loss": 1.1398,
+      "step": 216000
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.582989898989899e-05,
+      "loss": 1.14,
+      "step": 216500
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.581979797979798e-05,
+      "loss": 1.1376,
+      "step": 217000
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.5809717171717175e-05,
+      "loss": 1.1398,
+      "step": 217500
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.5799616161616164e-05,
+      "loss": 1.1377,
+      "step": 218000
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.5789515151515153e-05,
+      "loss": 1.1402,
+      "step": 218500
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.577941414141414e-05,
+      "loss": 1.1375,
+      "step": 219000
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 1.5769313131313133e-05,
+      "loss": 1.138,
+      "step": 219500
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 1.5759232323232324e-05,
+      "loss": 1.138,
+      "step": 220000
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 1.5749131313131316e-05,
+      "loss": 1.138,
+      "step": 220500
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.5739030303030304e-05,
+      "loss": 1.1397,
+      "step": 221000
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.5728929292929293e-05,
+      "loss": 1.1378,
+      "step": 221500
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.571886868686869e-05,
+      "loss": 1.1363,
+      "step": 222000
+    },
+    {
+      "epoch": 10.0,
+      "eval_accuracy": 0.7761830368284265,
+      "eval_loss": 0.9586696028709412,
+      "eval_runtime": 58.8614,
+      "eval_samples_per_second": 830.935,
+      "eval_steps_per_second": 12.997,
+      "step": 222020
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 1.5708767676767677e-05,
+      "loss": 1.1378,
+      "step": 222500
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 1.569866666666667e-05,
+      "loss": 1.1384,
+      "step": 223000
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 1.5688565656565658e-05,
+      "loss": 1.136,
+      "step": 223500
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 1.567848484848485e-05,
+      "loss": 1.136,
+      "step": 224000
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 1.566838383838384e-05,
+      "loss": 1.137,
+      "step": 224500
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 1.565828282828283e-05,
+      "loss": 1.1366,
+      "step": 225000
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 1.564818181818182e-05,
+      "loss": 1.1362,
+      "step": 225500
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 1.563808080808081e-05,
+      "loss": 1.1348,
+      "step": 226000
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 1.5627979797979798e-05,
+      "loss": 1.1345,
+      "step": 226500
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 1.561787878787879e-05,
+      "loss": 1.1347,
+      "step": 227000
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 1.560777777777778e-05,
+      "loss": 1.1378,
+      "step": 227500
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 1.5597676767676767e-05,
+      "loss": 1.1335,
+      "step": 228000
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 1.558759595959596e-05,
+      "loss": 1.1347,
+      "step": 228500
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 1.557749494949495e-05,
+      "loss": 1.1327,
+      "step": 229000
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 1.556739393939394e-05,
+      "loss": 1.1336,
+      "step": 229500
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 1.555729292929293e-05,
+      "loss": 1.1341,
+      "step": 230000
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 1.554719191919192e-05,
+      "loss": 1.1327,
+      "step": 230500
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 1.553709090909091e-05,
+      "loss": 1.1339,
+      "step": 231000
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 1.55269898989899e-05,
+      "loss": 1.1354,
+      "step": 231500
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 1.551688888888889e-05,
+      "loss": 1.1363,
+      "step": 232000
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 1.550678787878788e-05,
+      "loss": 1.1324,
+      "step": 232500
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 1.549670707070707e-05,
+      "loss": 1.1316,
+      "step": 233000
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 1.5486606060606063e-05,
+      "loss": 1.1336,
+      "step": 233500
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 1.547650505050505e-05,
+      "loss": 1.1309,
+      "step": 234000
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 1.5466404040404043e-05,
+      "loss": 1.1329,
+      "step": 234500
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 1.5456323232323234e-05,
+      "loss": 1.131,
+      "step": 235000
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 1.5446222222222222e-05,
+      "loss": 1.1321,
+      "step": 235500
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 1.5436121212121214e-05,
+      "loss": 1.132,
+      "step": 236000
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 1.5426040404040405e-05,
+      "loss": 1.1331,
+      "step": 236500
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 1.5415939393939394e-05,
+      "loss": 1.1303,
+      "step": 237000
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 1.5405838383838386e-05,
+      "loss": 1.1308,
+      "step": 237500
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 1.5395737373737378e-05,
+      "loss": 1.1311,
+      "step": 238000
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 1.5385636363636366e-05,
+      "loss": 1.1318,
+      "step": 238500
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 1.5375535353535355e-05,
+      "loss": 1.129,
+      "step": 239000
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 1.5365434343434343e-05,
+      "loss": 1.1309,
+      "step": 239500
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 1.5355333333333335e-05,
+      "loss": 1.1274,
+      "step": 240000
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 1.5345232323232324e-05,
+      "loss": 1.131,
+      "step": 240500
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 1.5335151515151518e-05,
+      "loss": 1.1312,
+      "step": 241000
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 1.5325050505050506e-05,
+      "loss": 1.1308,
+      "step": 241500
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 1.5314949494949495e-05,
+      "loss": 1.1275,
+      "step": 242000
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 1.5304848484848487e-05,
+      "loss": 1.1276,
+      "step": 242500
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 1.5294767676767678e-05,
+      "loss": 1.1279,
+      "step": 243000
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 1.5284666666666666e-05,
+      "loss": 1.1271,
+      "step": 243500
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 1.5274565656565658e-05,
+      "loss": 1.1298,
+      "step": 244000
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 0.7773125170940649,
+      "eval_loss": 0.9514342546463013,
+      "eval_runtime": 69.4149,
+      "eval_samples_per_second": 704.604,
+      "eval_steps_per_second": 11.021,
+      "step": 244222
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 1.5264464646464647e-05,
+      "loss": 1.1286,
+      "step": 244500
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 1.5254363636363639e-05,
+      "loss": 1.1277,
+      "step": 245000
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 1.5244282828282831e-05,
+      "loss": 1.1265,
+      "step": 245500
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 1.523418181818182e-05,
+      "loss": 1.13,
+      "step": 246000
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 1.522408080808081e-05,
+      "loss": 1.1257,
+      "step": 246500
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 1.5213979797979798e-05,
+      "loss": 1.1258,
+      "step": 247000
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 1.520387878787879e-05,
+      "loss": 1.1276,
+      "step": 247500
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 1.5193797979797981e-05,
+      "loss": 1.1293,
+      "step": 248000
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 1.5183696969696971e-05,
+      "loss": 1.1284,
+      "step": 248500
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 1.5173595959595962e-05,
+      "loss": 1.1292,
+      "step": 249000
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 1.516349494949495e-05,
+      "loss": 1.1233,
+      "step": 249500
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 1.515339393939394e-05,
+      "loss": 1.1245,
+      "step": 250000
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 1.5143292929292929e-05,
+      "loss": 1.1258,
+      "step": 250500
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 1.5133212121212121e-05,
+      "loss": 1.1235,
+      "step": 251000
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 1.5123111111111113e-05,
+      "loss": 1.1276,
+      "step": 251500
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 1.5113010101010102e-05,
+      "loss": 1.1251,
+      "step": 252000
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 1.5102909090909092e-05,
+      "loss": 1.1264,
+      "step": 252500
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 1.509280808080808e-05,
+      "loss": 1.1224,
+      "step": 253000
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 1.5082707070707073e-05,
+      "loss": 1.126,
+      "step": 253500
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 1.5072606060606061e-05,
+      "loss": 1.1239,
+      "step": 254000
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 1.5062525252525254e-05,
+      "loss": 1.1247,
+      "step": 254500
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 1.5052424242424244e-05,
+      "loss": 1.1228,
+      "step": 255000
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 1.5042343434343436e-05,
+      "loss": 1.1247,
+      "step": 255500
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 1.5032242424242425e-05,
+      "loss": 1.1251,
+      "step": 256000
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 1.5022141414141415e-05,
+      "loss": 1.1253,
+      "step": 256500
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 1.5012040404040405e-05,
+      "loss": 1.1245,
+      "step": 257000
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 1.5001939393939395e-05,
+      "loss": 1.123,
+      "step": 257500
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 1.4991838383838384e-05,
+      "loss": 1.123,
+      "step": 258000
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 1.4981737373737374e-05,
+      "loss": 1.125,
+      "step": 258500
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 1.4971636363636366e-05,
+      "loss": 1.1225,
+      "step": 259000
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 1.4961535353535355e-05,
+      "loss": 1.1244,
+      "step": 259500
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 1.4951434343434345e-05,
+      "loss": 1.1224,
+      "step": 260000
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 1.4941333333333334e-05,
+      "loss": 1.1227,
+      "step": 260500
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 1.4931232323232326e-05,
+      "loss": 1.1228,
+      "step": 261000
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 1.4921151515151518e-05,
+      "loss": 1.1221,
+      "step": 261500
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 1.4911050505050507e-05,
+      "loss": 1.1242,
+      "step": 262000
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 1.49009898989899e-05,
+      "loss": 1.1223,
+      "step": 262500
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 1.489088888888889e-05,
+      "loss": 1.1231,
+      "step": 263000
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 1.488078787878788e-05,
+      "loss": 1.1204,
+      "step": 263500
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 1.487068686868687e-05,
+      "loss": 1.1224,
+      "step": 264000
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 1.486058585858586e-05,
+      "loss": 1.1193,
+      "step": 264500
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 1.4850484848484849e-05,
+      "loss": 1.1209,
+      "step": 265000
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 1.4840383838383841e-05,
+      "loss": 1.1217,
+      "step": 265500
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 1.483028282828283e-05,
+      "loss": 1.1214,
+      "step": 266000
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.778167277137698,
+      "eval_loss": 0.9459095597267151,
+      "eval_runtime": 59.4353,
+      "eval_samples_per_second": 822.912,
+      "eval_steps_per_second": 12.871,
+      "step": 266424
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 1.482018181818182e-05,
+      "loss": 1.1203,
+      "step": 266500
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 1.4810101010101012e-05,
+      "loss": 1.1183,
+      "step": 267000
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 1.48e-05,
+      "loss": 1.1201,
+      "step": 267500
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 1.4789898989898993e-05,
+      "loss": 1.1192,
+      "step": 268000
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 1.4779797979797981e-05,
+      "loss": 1.1195,
+      "step": 268500
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 1.4769696969696971e-05,
+      "loss": 1.118,
+      "step": 269000
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 1.4759616161616164e-05,
+      "loss": 1.1212,
+      "step": 269500
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 1.4749515151515152e-05,
+      "loss": 1.117,
+      "step": 270000
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 1.4739414141414143e-05,
+      "loss": 1.1198,
+      "step": 270500
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 1.4729313131313133e-05,
+      "loss": 1.121,
+      "step": 271000
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 1.4719212121212123e-05,
+      "loss": 1.1198,
+      "step": 271500
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 1.4709111111111112e-05,
+      "loss": 1.1213,
+      "step": 272000
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 1.4699030303030304e-05,
+      "loss": 1.1188,
+      "step": 272500
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 1.4688949494949496e-05,
+      "loss": 1.12,
+      "step": 273000
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 1.4678848484848487e-05,
+      "loss": 1.1171,
+      "step": 273500
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 1.4668747474747475e-05,
+      "loss": 1.1191,
+      "step": 274000
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 1.4658646464646467e-05,
+      "loss": 1.1186,
+      "step": 274500
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 1.4648545454545456e-05,
+      "loss": 1.1186,
+      "step": 275000
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 1.4638444444444446e-05,
+      "loss": 1.1161,
+      "step": 275500
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 1.4628343434343435e-05,
+      "loss": 1.1188,
+      "step": 276000
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 1.4618242424242427e-05,
+      "loss": 1.1151,
+      "step": 276500
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 1.4608161616161619e-05,
+      "loss": 1.1159,
+      "step": 277000
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 1.4598060606060607e-05,
+      "loss": 1.1189,
+      "step": 277500
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 1.4587959595959598e-05,
+      "loss": 1.117,
+      "step": 278000
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 1.457787878787879e-05,
+      "loss": 1.1165,
+      "step": 278500
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 1.4567777777777779e-05,
+      "loss": 1.1193,
+      "step": 279000
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 1.4557676767676769e-05,
+      "loss": 1.1169,
+      "step": 279500
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 1.4547575757575757e-05,
+      "loss": 1.1169,
+      "step": 280000
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 1.453747474747475e-05,
+      "loss": 1.1173,
+      "step": 280500
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 1.4527373737373738e-05,
+      "loss": 1.1147,
+      "step": 281000
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 1.4517272727272728e-05,
+      "loss": 1.1163,
+      "step": 281500
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 1.4507171717171717e-05,
+      "loss": 1.1126,
+      "step": 282000
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 1.4497090909090909e-05,
+      "loss": 1.1145,
+      "step": 282500
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 1.4486989898989901e-05,
+      "loss": 1.1154,
+      "step": 283000
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 1.447688888888889e-05,
+      "loss": 1.1157,
+      "step": 283500
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 1.446678787878788e-05,
+      "loss": 1.1177,
+      "step": 284000
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 1.4456686868686869e-05,
+      "loss": 1.1137,
+      "step": 284500
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 1.444660606060606e-05,
+      "loss": 1.1135,
+      "step": 285000
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 1.4436525252525253e-05,
+      "loss": 1.1155,
+      "step": 285500
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 1.4426424242424243e-05,
+      "loss": 1.1157,
+      "step": 286000
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 1.4416323232323232e-05,
+      "loss": 1.1137,
+      "step": 286500
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 1.4406222222222224e-05,
+      "loss": 1.1127,
+      "step": 287000
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 1.4396121212121213e-05,
+      "loss": 1.1141,
+      "step": 287500
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 1.4386020202020203e-05,
+      "loss": 1.1135,
+      "step": 288000
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 1.4375919191919191e-05,
+      "loss": 1.115,
+      "step": 288500
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": 0.7789705199916636,
+      "eval_loss": 0.9406028985977173,
+      "eval_runtime": 53.5374,
+      "eval_samples_per_second": 913.567,
+      "eval_steps_per_second": 14.289,
+      "step": 288626
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 1.4365818181818183e-05,
+      "loss": 1.1135,
+      "step": 289000
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 1.4355717171717174e-05,
+      "loss": 1.115,
+      "step": 289500
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 1.4345636363636364e-05,
+      "loss": 1.1148,
+      "step": 290000
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 1.4335535353535355e-05,
+      "loss": 1.1116,
+      "step": 290500
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 1.4325434343434346e-05,
+      "loss": 1.1142,
+      "step": 291000
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 1.4315333333333335e-05,
+      "loss": 1.1112,
+      "step": 291500
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 1.4305232323232325e-05,
+      "loss": 1.1137,
+      "step": 292000
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 1.4295151515151516e-05,
+      "loss": 1.1116,
+      "step": 292500
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 1.4285050505050506e-05,
+      "loss": 1.1099,
+      "step": 293000
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 1.4274949494949496e-05,
+      "loss": 1.1123,
+      "step": 293500
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 1.4264848484848485e-05,
+      "loss": 1.111,
+      "step": 294000
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 1.4254747474747477e-05,
+      "loss": 1.1138,
+      "step": 294500
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 1.4244646464646466e-05,
+      "loss": 1.114,
+      "step": 295000
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 1.4234565656565658e-05,
+      "loss": 1.1103,
+      "step": 295500
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 1.4224464646464648e-05,
+      "loss": 1.1121,
+      "step": 296000
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 1.4214363636363637e-05,
+      "loss": 1.1105,
+      "step": 296500
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 1.4204262626262629e-05,
+      "loss": 1.1097,
+      "step": 297000
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 1.4194161616161617e-05,
+      "loss": 1.1137,
+      "step": 297500
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 1.4184060606060608e-05,
+      "loss": 1.1129,
+      "step": 298000
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 1.4173959595959596e-05,
+      "loss": 1.1074,
+      "step": 298500
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 1.4163898989898992e-05,
+      "loss": 1.1105,
+      "step": 299000
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 1.415379797979798e-05,
+      "loss": 1.1074,
+      "step": 299500
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 1.4143696969696971e-05,
+      "loss": 1.1093,
+      "step": 300000
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 1.413359595959596e-05,
+      "loss": 1.1067,
+      "step": 300500
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 1.4123494949494952e-05,
+      "loss": 1.1121,
+      "step": 301000
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 1.411339393939394e-05,
+      "loss": 1.1097,
+      "step": 301500
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 1.410329292929293e-05,
+      "loss": 1.1076,
+      "step": 302000
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 1.4093191919191919e-05,
+      "loss": 1.112,
+      "step": 302500
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 1.4083111111111111e-05,
+      "loss": 1.1113,
+      "step": 303000
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 1.4073010101010103e-05,
+      "loss": 1.1093,
+      "step": 303500
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 1.4062909090909092e-05,
+      "loss": 1.1126,
+      "step": 304000
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 1.4052808080808082e-05,
+      "loss": 1.1112,
+      "step": 304500
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 1.404270707070707e-05,
+      "loss": 1.1078,
+      "step": 305000
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 1.4032606060606063e-05,
+      "loss": 1.1097,
+      "step": 305500
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 1.4022505050505051e-05,
+      "loss": 1.1099,
+      "step": 306000
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 1.4012424242424244e-05,
+      "loss": 1.1086,
+      "step": 306500
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 1.4002323232323234e-05,
+      "loss": 1.109,
+      "step": 307000
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 1.3992222222222222e-05,
+      "loss": 1.1071,
+      "step": 307500
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 1.3982121212121213e-05,
+      "loss": 1.1077,
+      "step": 308000
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 1.3972040404040405e-05,
+      "loss": 1.1084,
+      "step": 308500
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 1.3961939393939394e-05,
+      "loss": 1.11,
+      "step": 309000
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 1.3951838383838386e-05,
+      "loss": 1.1084,
+      "step": 309500
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 1.3941737373737374e-05,
+      "loss": 1.107,
+      "step": 310000
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 1.3931656565656566e-05,
+      "loss": 1.1114,
+      "step": 310500
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.7802484831281742,
+      "eval_loss": 0.9347447752952576,
+      "eval_runtime": 54.177,
+      "eval_samples_per_second": 902.781,
+      "eval_steps_per_second": 14.12,
+      "step": 310828
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 1.3921555555555557e-05,
+      "loss": 1.1097,
+      "step": 311000
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 1.3911454545454545e-05,
+      "loss": 1.1045,
+      "step": 311500
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 1.3901353535353537e-05,
+      "loss": 1.108,
+      "step": 312000
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 1.3891252525252526e-05,
+      "loss": 1.1074,
+      "step": 312500
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 1.3881171717171718e-05,
+      "loss": 1.1066,
+      "step": 313000
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 1.3871070707070708e-05,
+      "loss": 1.1076,
+      "step": 313500
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 1.3860969696969697e-05,
+      "loss": 1.107,
+      "step": 314000
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 1.3850868686868689e-05,
+      "loss": 1.1064,
+      "step": 314500
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 1.3840767676767678e-05,
+      "loss": 1.1043,
+      "step": 315000
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 1.3830686868686868e-05,
+      "loss": 1.1044,
+      "step": 315500
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 1.382058585858586e-05,
+      "loss": 1.1053,
+      "step": 316000
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 1.3810505050505052e-05,
+      "loss": 1.1041,
+      "step": 316500
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 1.3800404040404041e-05,
+      "loss": 1.1079,
+      "step": 317000
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 1.3790303030303031e-05,
+      "loss": 1.1046,
+      "step": 317500
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 1.378020202020202e-05,
+      "loss": 1.1072,
+      "step": 318000
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 1.3770101010101012e-05,
+      "loss": 1.1059,
+      "step": 318500
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 1.376e-05,
+      "loss": 1.1062,
+      "step": 319000
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 1.374989898989899e-05,
+      "loss": 1.104,
+      "step": 319500
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 1.3739797979797983e-05,
+      "loss": 1.1037,
+      "step": 320000
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 1.3729696969696971e-05,
+      "loss": 1.1052,
+      "step": 320500
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 1.3719595959595962e-05,
+      "loss": 1.1044,
+      "step": 321000
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 1.370949494949495e-05,
+      "loss": 1.1051,
+      "step": 321500
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 1.3699414141414142e-05,
+      "loss": 1.1036,
+      "step": 322000
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 1.3689313131313133e-05,
+      "loss": 1.1043,
+      "step": 322500
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 1.3679212121212121e-05,
+      "loss": 1.1014,
+      "step": 323000
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 1.3669111111111113e-05,
+      "loss": 1.1032,
+      "step": 323500
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 1.3659030303030305e-05,
+      "loss": 1.1048,
+      "step": 324000
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 1.3648929292929294e-05,
+      "loss": 1.1034,
+      "step": 324500
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 1.3638828282828284e-05,
+      "loss": 1.1054,
+      "step": 325000
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 1.3628727272727273e-05,
+      "loss": 1.1044,
+      "step": 325500
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 1.3618626262626265e-05,
+      "loss": 1.1013,
+      "step": 326000
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 1.3608525252525254e-05,
+      "loss": 1.1056,
+      "step": 326500
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 1.3598424242424244e-05,
+      "loss": 1.1021,
+      "step": 327000
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 1.3588343434343436e-05,
+      "loss": 1.1018,
+      "step": 327500
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 1.3578242424242425e-05,
+      "loss": 1.1032,
+      "step": 328000
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 1.3568141414141417e-05,
+      "loss": 1.1047,
+      "step": 328500
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 1.3558040404040405e-05,
+      "loss": 1.1034,
+      "step": 329000
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 1.3547959595959597e-05,
+      "loss": 1.1012,
+      "step": 329500
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 1.3537878787878788e-05,
+      "loss": 1.1001,
+      "step": 330000
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 1.352777777777778e-05,
+      "loss": 1.1021,
+      "step": 330500
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 1.3517676767676769e-05,
+      "loss": 1.1035,
+      "step": 331000
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 1.3507575757575759e-05,
+      "loss": 1.1046,
+      "step": 331500
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 1.3497474747474747e-05,
+      "loss": 1.1032,
+      "step": 332000
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 1.348737373737374e-05,
+      "loss": 1.101,
+      "step": 332500
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 1.3477272727272728e-05,
+      "loss": 1.1021,
+      "step": 333000
+    },
+    {
+      "epoch": 15.0,
+      "eval_accuracy": 0.780922505853859,
+      "eval_loss": 0.9300829172134399,
+      "eval_runtime": 54.2448,
+      "eval_samples_per_second": 901.654,
+      "eval_steps_per_second": 14.103,
+      "step": 333030
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 1.3467171717171718e-05,
+      "loss": 1.101,
+      "step": 333500
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 1.3457070707070707e-05,
+      "loss": 1.0986,
+      "step": 334000
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 1.3446969696969699e-05,
+      "loss": 1.1012,
+      "step": 334500
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 1.3436888888888891e-05,
+      "loss": 1.1006,
+      "step": 335000
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 1.342678787878788e-05,
+      "loss": 1.0995,
+      "step": 335500
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 1.341668686868687e-05,
+      "loss": 1.1004,
+      "step": 336000
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 1.3406585858585859e-05,
+      "loss": 1.102,
+      "step": 336500
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 1.3396505050505051e-05,
+      "loss": 1.0979,
+      "step": 337000
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 1.3386404040404041e-05,
+      "loss": 1.099,
+      "step": 337500
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 1.337630303030303e-05,
+      "loss": 1.1008,
+      "step": 338000
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 1.3366202020202022e-05,
+      "loss": 1.0985,
+      "step": 338500
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 1.335610101010101e-05,
+      "loss": 1.097,
+      "step": 339000
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 1.3346e-05,
+      "loss": 1.1005,
+      "step": 339500
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 1.333589898989899e-05,
+      "loss": 1.0993,
+      "step": 340000
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 1.3325797979797981e-05,
+      "loss": 1.0975,
+      "step": 340500
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 1.3315696969696971e-05,
+      "loss": 1.0973,
+      "step": 341000
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 1.3305616161616162e-05,
+      "loss": 1.0995,
+      "step": 341500
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 1.3295515151515152e-05,
+      "loss": 1.1009,
+      "step": 342000
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 1.3285414141414144e-05,
+      "loss": 1.0983,
+      "step": 342500
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 1.3275313131313133e-05,
+      "loss": 1.0996,
+      "step": 343000
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 1.3265212121212123e-05,
+      "loss": 1.0998,
+      "step": 343500
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 1.3255131313131315e-05,
+      "loss": 1.0996,
+      "step": 344000
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 1.3245030303030304e-05,
+      "loss": 1.099,
+      "step": 344500
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 1.3234929292929294e-05,
+      "loss": 1.1004,
+      "step": 345000
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 1.3224828282828283e-05,
+      "loss": 1.0983,
+      "step": 345500
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 1.3214727272727275e-05,
+      "loss": 1.097,
+      "step": 346000
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 1.3204626262626263e-05,
+      "loss": 1.1022,
+      "step": 346500
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 1.3194545454545456e-05,
+      "loss": 1.0994,
+      "step": 347000
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 1.3184444444444446e-05,
+      "loss": 1.096,
+      "step": 347500
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 1.3174343434343435e-05,
+      "loss": 1.0994,
+      "step": 348000
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 1.3164242424242427e-05,
+      "loss": 1.0973,
+      "step": 348500
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 1.3154141414141415e-05,
+      "loss": 1.0969,
+      "step": 349000
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 1.3144060606060607e-05,
+      "loss": 1.0957,
+      "step": 349500
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 1.3133959595959598e-05,
+      "loss": 1.0978,
+      "step": 350000
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 1.3123858585858586e-05,
+      "loss": 1.0993,
+      "step": 350500
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 1.3113757575757578e-05,
+      "loss": 1.096,
+      "step": 351000
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 1.3103656565656567e-05,
+      "loss": 1.0964,
+      "step": 351500
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 1.3093555555555557e-05,
+      "loss": 1.0967,
+      "step": 352000
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 1.3083454545454546e-05,
+      "loss": 1.0973,
+      "step": 352500
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 1.3073353535353536e-05,
+      "loss": 1.0981,
+      "step": 353000
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 1.3063272727272728e-05,
+      "loss": 1.0963,
+      "step": 353500
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 1.305319191919192e-05,
+      "loss": 1.0956,
+      "step": 354000
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 1.3043111111111113e-05,
+      "loss": 1.0968,
+      "step": 354500
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 1.3033010101010101e-05,
+      "loss": 1.0974,
+      "step": 355000
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.7820052479123264,
+      "eval_loss": 0.9253054857254028,
+      "eval_runtime": 54.0729,
+      "eval_samples_per_second": 904.519,
+      "eval_steps_per_second": 14.148,
+      "step": 355232
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 1.3022909090909093e-05,
+      "loss": 1.0944,
+      "step": 355500
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 1.3012808080808082e-05,
+      "loss": 1.0943,
+      "step": 356000
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 1.3002707070707072e-05,
+      "loss": 1.0965,
+      "step": 356500
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 1.299260606060606e-05,
+      "loss": 1.0949,
+      "step": 357000
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 1.2982525252525253e-05,
+      "loss": 1.0929,
+      "step": 357500
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 1.2972424242424243e-05,
+      "loss": 1.0954,
+      "step": 358000
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 1.2962323232323234e-05,
+      "loss": 1.0964,
+      "step": 358500
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 1.2952222222222224e-05,
+      "loss": 1.0964,
+      "step": 359000
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 1.2942121212121213e-05,
+      "loss": 1.0961,
+      "step": 359500
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 1.2932020202020203e-05,
+      "loss": 1.0915,
+      "step": 360000
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 1.2921919191919191e-05,
+      "loss": 1.0916,
+      "step": 360500
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 1.2911818181818183e-05,
+      "loss": 1.0948,
+      "step": 361000
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 1.2901737373737376e-05,
+      "loss": 1.0927,
+      "step": 361500
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 1.2891636363636364e-05,
+      "loss": 1.0931,
+      "step": 362000
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 1.2881535353535355e-05,
+      "loss": 1.0941,
+      "step": 362500
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 1.2871454545454547e-05,
+      "loss": 1.0957,
+      "step": 363000
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 1.2861353535353535e-05,
+      "loss": 1.0926,
+      "step": 363500
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 1.2851252525252527e-05,
+      "loss": 1.0925,
+      "step": 364000
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 1.2841151515151516e-05,
+      "loss": 1.0941,
+      "step": 364500
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 1.2831050505050506e-05,
+      "loss": 1.0949,
+      "step": 365000
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 1.2820949494949495e-05,
+      "loss": 1.0926,
+      "step": 365500
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 1.2810848484848487e-05,
+      "loss": 1.0945,
+      "step": 366000
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 1.2800747474747475e-05,
+      "loss": 1.0934,
+      "step": 366500
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 1.2790646464646466e-05,
+      "loss": 1.0919,
+      "step": 367000
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 1.2780565656565658e-05,
+      "loss": 1.0924,
+      "step": 367500
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 1.2770464646464647e-05,
+      "loss": 1.0951,
+      "step": 368000
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 1.2760363636363637e-05,
+      "loss": 1.0931,
+      "step": 368500
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 1.2750262626262629e-05,
+      "loss": 1.0921,
+      "step": 369000
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 1.2740161616161617e-05,
+      "loss": 1.0918,
+      "step": 369500
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 1.2730060606060608e-05,
+      "loss": 1.092,
+      "step": 370000
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 1.2719959595959596e-05,
+      "loss": 1.0932,
+      "step": 370500
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 1.2709858585858588e-05,
+      "loss": 1.0927,
+      "step": 371000
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 1.269977777777778e-05,
+      "loss": 1.0927,
+      "step": 371500
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 1.2689676767676769e-05,
+      "loss": 1.0924,
+      "step": 372000
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 1.2679595959595961e-05,
+      "loss": 1.0925,
+      "step": 372500
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 1.2669494949494952e-05,
+      "loss": 1.0931,
+      "step": 373000
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 1.265939393939394e-05,
+      "loss": 1.0931,
+      "step": 373500
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 1.264929292929293e-05,
+      "loss": 1.0907,
+      "step": 374000
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 1.2639191919191919e-05,
+      "loss": 1.0895,
+      "step": 374500
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 1.2629111111111111e-05,
+      "loss": 1.0928,
+      "step": 375000
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 1.2619010101010103e-05,
+      "loss": 1.0919,
+      "step": 375500
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 1.2608909090909092e-05,
+      "loss": 1.0927,
+      "step": 376000
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 1.2598808080808082e-05,
+      "loss": 1.0934,
+      "step": 376500
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 1.258870707070707e-05,
+      "loss": 1.0907,
+      "step": 377000
+    },
+    {
+      "epoch": 17.0,
+      "eval_accuracy": 0.7825210862105197,
+      "eval_loss": 0.9215599298477173,
+      "eval_runtime": 53.7325,
+      "eval_samples_per_second": 910.25,
+      "eval_steps_per_second": 14.237,
+      "step": 377434
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 1.2578626262626263e-05,
+      "loss": 1.0906,
+      "step": 377500
+    },
+    {
+      "epoch": 17.03,
+      "learning_rate": 1.2568525252525255e-05,
+      "loss": 1.0915,
+      "step": 378000
+    },
+    {
+      "epoch": 17.05,
+      "learning_rate": 1.2558444444444447e-05,
+      "loss": 1.0909,
+      "step": 378500
+    },
+    {
+      "epoch": 17.07,
+      "learning_rate": 1.2548343434343436e-05,
+      "loss": 1.0899,
+      "step": 379000
+    },
+    {
+      "epoch": 17.09,
+      "learning_rate": 1.2538242424242426e-05,
+      "loss": 1.0899,
+      "step": 379500
+    },
+    {
+      "epoch": 17.12,
+      "learning_rate": 1.2528141414141415e-05,
+      "loss": 1.0896,
+      "step": 380000
+    },
+    {
+      "epoch": 17.14,
+      "learning_rate": 1.2518040404040405e-05,
+      "loss": 1.0891,
+      "step": 380500
+    },
+    {
+      "epoch": 17.16,
+      "learning_rate": 1.2507939393939395e-05,
+      "loss": 1.0892,
+      "step": 381000
+    },
+    {
+      "epoch": 17.18,
+      "learning_rate": 1.2497858585858586e-05,
+      "loss": 1.0886,
+      "step": 381500
+    },
+    {
+      "epoch": 17.21,
+      "learning_rate": 1.2487757575757578e-05,
+      "loss": 1.0934,
+      "step": 382000
+    },
+    {
+      "epoch": 17.23,
+      "learning_rate": 1.2477656565656566e-05,
+      "loss": 1.0887,
+      "step": 382500
+    },
+    {
+      "epoch": 17.25,
+      "learning_rate": 1.2467555555555557e-05,
+      "loss": 1.0898,
+      "step": 383000
+    },
+    {
+      "epoch": 17.27,
+      "learning_rate": 1.2457454545454545e-05,
+      "loss": 1.0899,
+      "step": 383500
+    },
+    {
+      "epoch": 17.3,
+      "learning_rate": 1.2447373737373738e-05,
+      "loss": 1.0886,
+      "step": 384000
+    },
+    {
+      "epoch": 17.32,
+      "learning_rate": 1.243727272727273e-05,
+      "loss": 1.0893,
+      "step": 384500
+    },
+    {
+      "epoch": 17.34,
+      "learning_rate": 1.2427171717171718e-05,
+      "loss": 1.0895,
+      "step": 385000
+    },
+    {
+      "epoch": 17.36,
+      "learning_rate": 1.2417070707070708e-05,
+      "loss": 1.0902,
+      "step": 385500
+    },
+    {
+      "epoch": 17.39,
+      "learning_rate": 1.2406969696969697e-05,
+      "loss": 1.0886,
+      "step": 386000
+    },
+    {
+      "epoch": 17.41,
+      "learning_rate": 1.2396868686868689e-05,
+      "loss": 1.0895,
+      "step": 386500
+    },
+    {
+      "epoch": 17.43,
+      "learning_rate": 1.2386767676767678e-05,
+      "loss": 1.0889,
+      "step": 387000
+    },
+    {
+      "epoch": 17.45,
+      "learning_rate": 1.2376666666666668e-05,
+      "loss": 1.0903,
+      "step": 387500
+    },
+    {
+      "epoch": 17.48,
+      "learning_rate": 1.2366565656565656e-05,
+      "loss": 1.0899,
+      "step": 388000
+    },
+    {
+      "epoch": 17.5,
+      "learning_rate": 1.2356505050505052e-05,
+      "loss": 1.0882,
+      "step": 388500
+    },
+    {
+      "epoch": 17.52,
+      "learning_rate": 1.2346404040404041e-05,
+      "loss": 1.0897,
+      "step": 389000
+    },
+    {
+      "epoch": 17.54,
+      "learning_rate": 1.2336303030303031e-05,
+      "loss": 1.0888,
+      "step": 389500
+    },
+    {
+      "epoch": 17.57,
+      "learning_rate": 1.232620202020202e-05,
+      "loss": 1.0874,
+      "step": 390000
+    },
+    {
+      "epoch": 17.59,
+      "learning_rate": 1.2316101010101012e-05,
+      "loss": 1.0881,
+      "step": 390500
+    },
+    {
+      "epoch": 17.61,
+      "learning_rate": 1.2306020202020204e-05,
+      "loss": 1.0902,
+      "step": 391000
+    },
+    {
+      "epoch": 17.63,
+      "learning_rate": 1.2295919191919193e-05,
+      "loss": 1.0853,
+      "step": 391500
+    },
+    {
+      "epoch": 17.66,
+      "learning_rate": 1.2285818181818183e-05,
+      "loss": 1.0914,
+      "step": 392000
+    },
+    {
+      "epoch": 17.68,
+      "learning_rate": 1.2275717171717172e-05,
+      "loss": 1.0886,
+      "step": 392500
+    },
+    {
+      "epoch": 17.7,
+      "learning_rate": 1.2265616161616164e-05,
+      "loss": 1.0892,
+      "step": 393000
+    },
+    {
+      "epoch": 17.72,
+      "learning_rate": 1.2255515151515152e-05,
+      "loss": 1.0896,
+      "step": 393500
+    },
+    {
+      "epoch": 17.75,
+      "learning_rate": 1.2245414141414142e-05,
+      "loss": 1.0896,
+      "step": 394000
+    },
+    {
+      "epoch": 17.77,
+      "learning_rate": 1.2235313131313131e-05,
+      "loss": 1.0855,
+      "step": 394500
+    },
+    {
+      "epoch": 17.79,
+      "learning_rate": 1.2225212121212123e-05,
+      "loss": 1.0861,
+      "step": 395000
+    },
+    {
+      "epoch": 17.81,
+      "learning_rate": 1.2215111111111113e-05,
+      "loss": 1.0858,
+      "step": 395500
+    },
+    {
+      "epoch": 17.84,
+      "learning_rate": 1.2205010101010102e-05,
+      "loss": 1.0848,
+      "step": 396000
+    },
+    {
+      "epoch": 17.86,
+      "learning_rate": 1.2194929292929294e-05,
+      "loss": 1.0887,
+      "step": 396500
+    },
+    {
+      "epoch": 17.88,
+      "learning_rate": 1.2184828282828284e-05,
+      "loss": 1.0889,
+      "step": 397000
+    },
+    {
+      "epoch": 17.9,
+      "learning_rate": 1.2174727272727273e-05,
+      "loss": 1.0884,
+      "step": 397500
+    },
+    {
+      "epoch": 17.93,
+      "learning_rate": 1.2164626262626265e-05,
+      "loss": 1.0895,
+      "step": 398000
+    },
+    {
+      "epoch": 17.95,
+      "learning_rate": 1.2154525252525254e-05,
+      "loss": 1.0864,
+      "step": 398500
+    },
+    {
+      "epoch": 17.97,
+      "learning_rate": 1.2144444444444446e-05,
+      "loss": 1.0863,
+      "step": 399000
+    },
+    {
+      "epoch": 17.99,
+      "learning_rate": 1.2134343434343436e-05,
+      "loss": 1.0847,
+      "step": 399500
+    },
+    {
+      "epoch": 18.0,
+      "eval_accuracy": 0.7831444451406563,
+      "eval_loss": 0.9177064895629883,
+      "eval_runtime": 53.7489,
+      "eval_samples_per_second": 909.972,
+      "eval_steps_per_second": 14.233,
+      "step": 399636
+    },
+    {
+      "epoch": 18.02,
+      "learning_rate": 1.2124242424242425e-05,
+      "loss": 1.0856,
+      "step": 400000
+    },
+    {
+      "epoch": 18.04,
+      "learning_rate": 1.2114141414141417e-05,
+      "loss": 1.0834,
+      "step": 400500
+    },
+    {
+      "epoch": 18.06,
+      "learning_rate": 1.2104060606060607e-05,
+      "loss": 1.0878,
+      "step": 401000
+    },
+    {
+      "epoch": 18.08,
+      "learning_rate": 1.2093959595959598e-05,
+      "loss": 1.0847,
+      "step": 401500
+    },
+    {
+      "epoch": 18.11,
+      "learning_rate": 1.2083858585858588e-05,
+      "loss": 1.0831,
+      "step": 402000
+    },
+    {
+      "epoch": 18.13,
+      "learning_rate": 1.2073757575757576e-05,
+      "loss": 1.0866,
+      "step": 402500
+    },
+    {
+      "epoch": 18.15,
+      "learning_rate": 1.2063656565656567e-05,
+      "loss": 1.0866,
+      "step": 403000
+    },
+    {
+      "epoch": 18.17,
+      "learning_rate": 1.2053555555555557e-05,
+      "loss": 1.0883,
+      "step": 403500
+    },
+    {
+      "epoch": 18.2,
+      "learning_rate": 1.2043454545454547e-05,
+      "loss": 1.0859,
+      "step": 404000
+    },
+    {
+      "epoch": 18.22,
+      "learning_rate": 1.2033353535353536e-05,
+      "loss": 1.0871,
+      "step": 404500
+    },
+    {
+      "epoch": 18.24,
+      "learning_rate": 1.2023252525252526e-05,
+      "loss": 1.0869,
+      "step": 405000
+    },
+    {
+      "epoch": 18.26,
+      "learning_rate": 1.2013151515151515e-05,
+      "loss": 1.0851,
+      "step": 405500
+    },
+    {
+      "epoch": 18.29,
+      "learning_rate": 1.2003050505050507e-05,
+      "loss": 1.0848,
+      "step": 406000
+    },
+    {
+      "epoch": 18.31,
+      "learning_rate": 1.1992949494949495e-05,
+      "loss": 1.0812,
+      "step": 406500
+    },
+    {
+      "epoch": 18.33,
+      "learning_rate": 1.1982868686868688e-05,
+      "loss": 1.0836,
+      "step": 407000
+    },
+    {
+      "epoch": 18.35,
+      "learning_rate": 1.1972767676767678e-05,
+      "loss": 1.0827,
+      "step": 407500
+    },
+    {
+      "epoch": 18.38,
+      "learning_rate": 1.1962666666666666e-05,
+      "loss": 1.0865,
+      "step": 408000
+    },
+    {
+      "epoch": 18.4,
+      "learning_rate": 1.1952585858585859e-05,
+      "loss": 1.0842,
+      "step": 408500
+    },
+    {
+      "epoch": 18.42,
+      "learning_rate": 1.194248484848485e-05,
+      "loss": 1.0835,
+      "step": 409000
+    },
+    {
+      "epoch": 18.44,
+      "learning_rate": 1.1932404040404041e-05,
+      "loss": 1.088,
+      "step": 409500
+    },
+    {
+      "epoch": 18.47,
+      "learning_rate": 1.1922303030303031e-05,
+      "loss": 1.0823,
+      "step": 410000
+    },
+    {
+      "epoch": 18.49,
+      "learning_rate": 1.1912202020202022e-05,
+      "loss": 1.0819,
+      "step": 410500
+    },
+    {
+      "epoch": 18.51,
+      "learning_rate": 1.190210101010101e-05,
+      "loss": 1.0859,
+      "step": 411000
+    },
+    {
+      "epoch": 18.53,
+      "learning_rate": 1.1892020202020203e-05,
+      "loss": 1.0867,
+      "step": 411500
+    },
+    {
+      "epoch": 18.56,
+      "learning_rate": 1.1881919191919193e-05,
+      "loss": 1.0836,
+      "step": 412000
+    },
+    {
+      "epoch": 18.58,
+      "learning_rate": 1.1871818181818182e-05,
+      "loss": 1.0846,
+      "step": 412500
+    },
+    {
+      "epoch": 18.6,
+      "learning_rate": 1.1861717171717173e-05,
+      "loss": 1.0861,
+      "step": 413000
+    },
+    {
+      "epoch": 18.62,
+      "learning_rate": 1.1851616161616162e-05,
+      "loss": 1.0863,
+      "step": 413500
+    },
+    {
+      "epoch": 18.65,
+      "learning_rate": 1.1841515151515152e-05,
+      "loss": 1.0851,
+      "step": 414000
+    },
+    {
+      "epoch": 18.67,
+      "learning_rate": 1.1831414141414141e-05,
+      "loss": 1.0832,
+      "step": 414500
+    },
+    {
+      "epoch": 18.69,
+      "learning_rate": 1.1821313131313133e-05,
+      "loss": 1.0823,
+      "step": 415000
+    },
+    {
+      "epoch": 18.71,
+      "learning_rate": 1.1811212121212122e-05,
+      "loss": 1.0847,
+      "step": 415500
+    },
+    {
+      "epoch": 18.74,
+      "learning_rate": 1.1801111111111112e-05,
+      "loss": 1.0851,
+      "step": 416000
+    },
+    {
+      "epoch": 18.76,
+      "learning_rate": 1.1791010101010104e-05,
+      "loss": 1.0828,
+      "step": 416500
+    },
+    {
+      "epoch": 18.78,
+      "learning_rate": 1.1780909090909092e-05,
+      "loss": 1.0832,
+      "step": 417000
+    },
+    {
+      "epoch": 18.8,
+      "learning_rate": 1.1770828282828285e-05,
+      "loss": 1.0848,
+      "step": 417500
+    },
+    {
+      "epoch": 18.83,
+      "learning_rate": 1.1760727272727275e-05,
+      "loss": 1.0839,
+      "step": 418000
+    },
+    {
+      "epoch": 18.85,
+      "learning_rate": 1.1750626262626264e-05,
+      "loss": 1.0854,
+      "step": 418500
+    },
+    {
+      "epoch": 18.87,
+      "learning_rate": 1.1740525252525254e-05,
+      "loss": 1.0878,
+      "step": 419000
+    },
+    {
+      "epoch": 18.89,
+      "learning_rate": 1.1730424242424242e-05,
+      "loss": 1.0803,
+      "step": 419500
+    },
+    {
+      "epoch": 18.92,
+      "learning_rate": 1.1720343434343435e-05,
+      "loss": 1.0815,
+      "step": 420000
+    },
+    {
+      "epoch": 18.94,
+      "learning_rate": 1.1710242424242427e-05,
+      "loss": 1.0809,
+      "step": 420500
+    },
+    {
+      "epoch": 18.96,
+      "learning_rate": 1.1700141414141415e-05,
+      "loss": 1.0812,
+      "step": 421000
+    },
+    {
+      "epoch": 18.98,
+      "learning_rate": 1.1690040404040406e-05,
+      "loss": 1.0815,
+      "step": 421500
+    },
+    {
+      "epoch": 19.0,
+      "eval_accuracy": 0.7838057953249593,
+      "eval_loss": 0.9133327007293701,
+      "eval_runtime": 53.8388,
+      "eval_samples_per_second": 908.453,
+      "eval_steps_per_second": 14.209,
+      "step": 421838
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 1.1679959595959598e-05,
+      "loss": 1.0823,
+      "step": 422000
+    },
+    {
+      "epoch": 19.03,
+      "learning_rate": 1.1669858585858586e-05,
+      "loss": 1.0791,
+      "step": 422500
+    },
+    {
+      "epoch": 19.05,
+      "learning_rate": 1.1659757575757578e-05,
+      "loss": 1.08,
+      "step": 423000
+    },
+    {
+      "epoch": 19.07,
+      "learning_rate": 1.164969696969697e-05,
+      "loss": 1.0812,
+      "step": 423500
+    },
+    {
+      "epoch": 19.1,
+      "learning_rate": 1.163959595959596e-05,
+      "loss": 1.08,
+      "step": 424000
+    },
+    {
+      "epoch": 19.12,
+      "learning_rate": 1.162949494949495e-05,
+      "loss": 1.0828,
+      "step": 424500
+    },
+    {
+      "epoch": 19.14,
+      "learning_rate": 1.1619393939393942e-05,
+      "loss": 1.0801,
+      "step": 425000
+    },
+    {
+      "epoch": 19.16,
+      "learning_rate": 1.160929292929293e-05,
+      "loss": 1.0807,
+      "step": 425500
+    },
+    {
+      "epoch": 19.19,
+      "learning_rate": 1.159919191919192e-05,
+      "loss": 1.0822,
+      "step": 426000
+    },
+    {
+      "epoch": 19.21,
+      "learning_rate": 1.158909090909091e-05,
+      "loss": 1.08,
+      "step": 426500
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 1.1578989898989901e-05,
+      "loss": 1.0822,
+      "step": 427000
+    },
+    {
+      "epoch": 19.26,
+      "learning_rate": 1.1568909090909093e-05,
+      "loss": 1.0818,
+      "step": 427500
+    },
+    {
+      "epoch": 19.28,
+      "learning_rate": 1.1558828282828282e-05,
+      "loss": 1.0797,
+      "step": 428000
+    },
+    {
+      "epoch": 19.3,
+      "learning_rate": 1.1548727272727274e-05,
+      "loss": 1.0823,
+      "step": 428500
+    },
+    {
+      "epoch": 19.32,
+      "learning_rate": 1.1538626262626265e-05,
+      "loss": 1.0825,
+      "step": 429000
+    },
+    {
+      "epoch": 19.35,
+      "learning_rate": 1.1528525252525253e-05,
+      "loss": 1.0785,
+      "step": 429500
+    },
+    {
+      "epoch": 19.37,
+      "learning_rate": 1.1518424242424245e-05,
+      "loss": 1.0795,
+      "step": 430000
+    },
+    {
+      "epoch": 19.39,
+      "learning_rate": 1.1508323232323234e-05,
+      "loss": 1.0818,
+      "step": 430500
+    },
+    {
+      "epoch": 19.41,
+      "learning_rate": 1.1498222222222224e-05,
+      "loss": 1.0818,
+      "step": 431000
+    },
+    {
+      "epoch": 19.44,
+      "learning_rate": 1.1488121212121213e-05,
+      "loss": 1.081,
+      "step": 431500
+    },
+    {
+      "epoch": 19.46,
+      "learning_rate": 1.1478020202020203e-05,
+      "loss": 1.0801,
+      "step": 432000
+    },
+    {
+      "epoch": 19.48,
+      "learning_rate": 1.1467939393939395e-05,
+      "loss": 1.0832,
+      "step": 432500
+    },
+    {
+      "epoch": 19.5,
+      "learning_rate": 1.1457838383838384e-05,
+      "loss": 1.0823,
+      "step": 433000
+    },
+    {
+      "epoch": 19.53,
+      "learning_rate": 1.1447737373737376e-05,
+      "loss": 1.0809,
+      "step": 433500
+    },
+    {
+      "epoch": 19.55,
+      "learning_rate": 1.1437636363636364e-05,
+      "loss": 1.0788,
+      "step": 434000
+    },
+    {
+      "epoch": 19.57,
+      "learning_rate": 1.1427555555555557e-05,
+      "loss": 1.0799,
+      "step": 434500
+    },
+    {
+      "epoch": 19.59,
+      "learning_rate": 1.1417454545454547e-05,
+      "loss": 1.0816,
+      "step": 435000
+    },
+    {
+      "epoch": 19.62,
+      "learning_rate": 1.1407353535353535e-05,
+      "loss": 1.0784,
+      "step": 435500
+    },
+    {
+      "epoch": 19.64,
+      "learning_rate": 1.1397272727272728e-05,
+      "loss": 1.0803,
+      "step": 436000
+    },
+    {
+      "epoch": 19.66,
+      "learning_rate": 1.138717171717172e-05,
+      "loss": 1.0776,
+      "step": 436500
+    },
+    {
+      "epoch": 19.68,
+      "learning_rate": 1.1377070707070708e-05,
+      "loss": 1.0799,
+      "step": 437000
+    },
+    {
+      "epoch": 19.71,
+      "learning_rate": 1.1366969696969699e-05,
+      "loss": 1.0806,
+      "step": 437500
+    },
+    {
+      "epoch": 19.73,
+      "learning_rate": 1.1356868686868687e-05,
+      "loss": 1.0792,
+      "step": 438000
+    },
+    {
+      "epoch": 19.75,
+      "learning_rate": 1.1346767676767677e-05,
+      "loss": 1.0782,
+      "step": 438500
+    },
+    {
+      "epoch": 19.77,
+      "learning_rate": 1.1336666666666668e-05,
+      "loss": 1.0774,
+      "step": 439000
+    },
+    {
+      "epoch": 19.8,
+      "learning_rate": 1.1326565656565658e-05,
+      "loss": 1.077,
+      "step": 439500
+    },
+    {
+      "epoch": 19.82,
+      "learning_rate": 1.1316464646464647e-05,
+      "loss": 1.0789,
+      "step": 440000
+    },
+    {
+      "epoch": 19.84,
+      "learning_rate": 1.1306404040404042e-05,
+      "loss": 1.0766,
+      "step": 440500
+    },
+    {
+      "epoch": 19.86,
+      "learning_rate": 1.1296303030303031e-05,
+      "loss": 1.0794,
+      "step": 441000
+    },
+    {
+      "epoch": 19.89,
+      "learning_rate": 1.1286202020202021e-05,
+      "loss": 1.0774,
+      "step": 441500
+    },
+    {
+      "epoch": 19.91,
+      "learning_rate": 1.127610101010101e-05,
+      "loss": 1.0788,
+      "step": 442000
+    },
+    {
+      "epoch": 19.93,
+      "learning_rate": 1.1266000000000002e-05,
+      "loss": 1.0821,
+      "step": 442500
+    },
+    {
+      "epoch": 19.95,
+      "learning_rate": 1.125589898989899e-05,
+      "loss": 1.0791,
+      "step": 443000
+    },
+    {
+      "epoch": 19.98,
+      "learning_rate": 1.124579797979798e-05,
+      "loss": 1.0788,
+      "step": 443500
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 1.1235717171717173e-05,
+      "loss": 1.0776,
+      "step": 444000
+    },
+    {
+      "epoch": 20.0,
+      "eval_accuracy": 0.783883000470097,
+      "eval_loss": 0.9127400517463684,
+      "eval_runtime": 53.9539,
+      "eval_samples_per_second": 906.515,
+      "eval_steps_per_second": 14.179,
+      "step": 444040
+    },
+    {
+      "epoch": 20.02,
+      "learning_rate": 1.1225616161616162e-05,
+      "loss": 1.0773,
+      "step": 444500
+    },
+    {
+      "epoch": 20.04,
+      "learning_rate": 1.1215515151515154e-05,
+      "loss": 1.0779,
+      "step": 445000
+    },
+    {
+      "epoch": 20.07,
+      "learning_rate": 1.1205414141414142e-05,
+      "loss": 1.0793,
+      "step": 445500
+    },
+    {
+      "epoch": 20.09,
+      "learning_rate": 1.1195313131313132e-05,
+      "loss": 1.0772,
+      "step": 446000
+    },
+    {
+      "epoch": 20.11,
+      "learning_rate": 1.1185212121212121e-05,
+      "loss": 1.0788,
+      "step": 446500
+    },
+    {
+      "epoch": 20.13,
+      "learning_rate": 1.1175111111111111e-05,
+      "loss": 1.077,
+      "step": 447000
+    },
+    {
+      "epoch": 20.16,
+      "learning_rate": 1.1165010101010102e-05,
+      "loss": 1.0774,
+      "step": 447500
+    },
+    {
+      "epoch": 20.18,
+      "learning_rate": 1.1154909090909092e-05,
+      "loss": 1.0766,
+      "step": 448000
+    },
+    {
+      "epoch": 20.2,
+      "learning_rate": 1.1144808080808082e-05,
+      "loss": 1.0815,
+      "step": 448500
+    },
+    {
+      "epoch": 20.22,
+      "learning_rate": 1.113470707070707e-05,
+      "loss": 1.0749,
+      "step": 449000
+    },
+    {
+      "epoch": 20.25,
+      "learning_rate": 1.1124626262626263e-05,
+      "loss": 1.079,
+      "step": 449500
+    },
+    {
+      "epoch": 20.27,
+      "learning_rate": 1.1114525252525255e-05,
+      "loss": 1.0772,
+      "step": 450000
+    },
+    {
+      "epoch": 20.29,
+      "learning_rate": 1.1104424242424244e-05,
+      "loss": 1.0774,
+      "step": 450500
+    },
+    {
+      "epoch": 20.31,
+      "learning_rate": 1.1094323232323234e-05,
+      "loss": 1.0804,
+      "step": 451000
+    },
+    {
+      "epoch": 20.34,
+      "learning_rate": 1.1084242424242426e-05,
+      "loss": 1.078,
+      "step": 451500
+    },
+    {
+      "epoch": 20.36,
+      "learning_rate": 1.1074161616161617e-05,
+      "loss": 1.0768,
+      "step": 452000
+    },
+    {
+      "epoch": 20.38,
+      "learning_rate": 1.1064060606060607e-05,
+      "loss": 1.0759,
+      "step": 452500
+    },
+    {
+      "epoch": 20.4,
+      "learning_rate": 1.1053959595959596e-05,
+      "loss": 1.0754,
+      "step": 453000
+    },
+    {
+      "epoch": 20.43,
+      "learning_rate": 1.1043858585858586e-05,
+      "loss": 1.0768,
+      "step": 453500
+    },
+    {
+      "epoch": 20.45,
+      "learning_rate": 1.1033777777777778e-05,
+      "loss": 1.0755,
+      "step": 454000
+    },
+    {
+      "epoch": 20.47,
+      "learning_rate": 1.1023676767676767e-05,
+      "loss": 1.0762,
+      "step": 454500
+    },
+    {
+      "epoch": 20.49,
+      "learning_rate": 1.1013575757575759e-05,
+      "loss": 1.0769,
+      "step": 455000
+    },
+    {
+      "epoch": 20.52,
+      "learning_rate": 1.1003474747474749e-05,
+      "loss": 1.074,
+      "step": 455500
+    },
+    {
+      "epoch": 20.54,
+      "learning_rate": 1.0993373737373738e-05,
+      "loss": 1.0772,
+      "step": 456000
+    },
+    {
+      "epoch": 20.56,
+      "learning_rate": 1.098327272727273e-05,
+      "loss": 1.0775,
+      "step": 456500
+    },
+    {
+      "epoch": 20.58,
+      "learning_rate": 1.0973171717171718e-05,
+      "loss": 1.0775,
+      "step": 457000
+    },
+    {
+      "epoch": 20.61,
+      "learning_rate": 1.0963070707070708e-05,
+      "loss": 1.0772,
+      "step": 457500
+    },
+    {
+      "epoch": 20.63,
+      "learning_rate": 1.09529898989899e-05,
+      "loss": 1.0761,
+      "step": 458000
+    },
+    {
+      "epoch": 20.65,
+      "learning_rate": 1.094288888888889e-05,
+      "loss": 1.076,
+      "step": 458500
+    },
+    {
+      "epoch": 20.67,
+      "learning_rate": 1.0932787878787881e-05,
+      "loss": 1.0753,
+      "step": 459000
+    },
+    {
+      "epoch": 20.7,
+      "learning_rate": 1.092268686868687e-05,
+      "loss": 1.078,
+      "step": 459500
+    },
+    {
+      "epoch": 20.72,
+      "learning_rate": 1.091258585858586e-05,
+      "loss": 1.0759,
+      "step": 460000
+    },
+    {
+      "epoch": 20.74,
+      "learning_rate": 1.0902484848484849e-05,
+      "loss": 1.0764,
+      "step": 460500
+    },
+    {
+      "epoch": 20.76,
+      "learning_rate": 1.0892383838383839e-05,
+      "loss": 1.076,
+      "step": 461000
+    },
+    {
+      "epoch": 20.79,
+      "learning_rate": 1.0882303030303031e-05,
+      "loss": 1.0756,
+      "step": 461500
+    },
+    {
+      "epoch": 20.81,
+      "learning_rate": 1.087220202020202e-05,
+      "loss": 1.0749,
+      "step": 462000
+    },
+    {
+      "epoch": 20.83,
+      "learning_rate": 1.0862101010101012e-05,
+      "loss": 1.0755,
+      "step": 462500
+    },
+    {
+      "epoch": 20.85,
+      "learning_rate": 1.0852e-05,
+      "loss": 1.078,
+      "step": 463000
+    },
+    {
+      "epoch": 20.88,
+      "learning_rate": 1.0841919191919193e-05,
+      "loss": 1.0725,
+      "step": 463500
+    },
+    {
+      "epoch": 20.9,
+      "learning_rate": 1.0831818181818183e-05,
+      "loss": 1.0725,
+      "step": 464000
+    },
+    {
+      "epoch": 20.92,
+      "learning_rate": 1.0821717171717172e-05,
+      "loss": 1.0756,
+      "step": 464500
+    },
+    {
+      "epoch": 20.94,
+      "learning_rate": 1.0811616161616164e-05,
+      "loss": 1.0754,
+      "step": 465000
+    },
+    {
+      "epoch": 20.97,
+      "learning_rate": 1.0801515151515152e-05,
+      "loss": 1.0766,
+      "step": 465500
+    },
+    {
+      "epoch": 20.99,
+      "learning_rate": 1.0791414141414142e-05,
+      "loss": 1.0755,
+      "step": 466000
+    },
+    {
+      "epoch": 21.0,
+      "eval_accuracy": 0.784943745573227,
+      "eval_loss": 0.9083593487739563,
+      "eval_runtime": 54.0471,
+      "eval_samples_per_second": 904.951,
+      "eval_steps_per_second": 14.154,
+      "step": 466242
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 1.0781313131313131e-05,
+      "loss": 1.0746,
+      "step": 466500
+    },
+    {
+      "epoch": 21.03,
+      "learning_rate": 1.0771212121212123e-05,
+      "loss": 1.0744,
+      "step": 467000
+    },
+    {
+      "epoch": 21.06,
+      "learning_rate": 1.0761111111111112e-05,
+      "loss": 1.0733,
+      "step": 467500
+    },
+    {
+      "epoch": 21.08,
+      "learning_rate": 1.0751010101010102e-05,
+      "loss": 1.073,
+      "step": 468000
+    },
+    {
+      "epoch": 21.1,
+      "learning_rate": 1.074090909090909e-05,
+      "loss": 1.0748,
+      "step": 468500
+    },
+    {
+      "epoch": 21.12,
+      "learning_rate": 1.0730848484848486e-05,
+      "loss": 1.0721,
+      "step": 469000
+    },
+    {
+      "epoch": 21.15,
+      "learning_rate": 1.0720747474747475e-05,
+      "loss": 1.0777,
+      "step": 469500
+    },
+    {
+      "epoch": 21.17,
+      "learning_rate": 1.0710646464646465e-05,
+      "loss": 1.0733,
+      "step": 470000
+    },
+    {
+      "epoch": 21.19,
+      "learning_rate": 1.0700545454545454e-05,
+      "loss": 1.072,
+      "step": 470500
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 1.0690444444444446e-05,
+      "loss": 1.0708,
+      "step": 471000
+    },
+    {
+      "epoch": 21.24,
+      "learning_rate": 1.0680343434343434e-05,
+      "loss": 1.0723,
+      "step": 471500
+    },
+    {
+      "epoch": 21.26,
+      "learning_rate": 1.0670242424242425e-05,
+      "loss": 1.0741,
+      "step": 472000
+    },
+    {
+      "epoch": 21.28,
+      "learning_rate": 1.0660141414141417e-05,
+      "loss": 1.0743,
+      "step": 472500
+    },
+    {
+      "epoch": 21.3,
+      "learning_rate": 1.0650040404040405e-05,
+      "loss": 1.0692,
+      "step": 473000
+    },
+    {
+      "epoch": 21.33,
+      "learning_rate": 1.0639959595959598e-05,
+      "loss": 1.0738,
+      "step": 473500
+    },
+    {
+      "epoch": 21.35,
+      "learning_rate": 1.0629858585858586e-05,
+      "loss": 1.0738,
+      "step": 474000
+    },
+    {
+      "epoch": 21.37,
+      "learning_rate": 1.0619757575757576e-05,
+      "loss": 1.0729,
+      "step": 474500
+    },
+    {
+      "epoch": 21.39,
+      "learning_rate": 1.0609656565656567e-05,
+      "loss": 1.0747,
+      "step": 475000
+    },
+    {
+      "epoch": 21.42,
+      "learning_rate": 1.0599555555555557e-05,
+      "loss": 1.0727,
+      "step": 475500
+    },
+    {
+      "epoch": 21.44,
+      "learning_rate": 1.0589454545454547e-05,
+      "loss": 1.0747,
+      "step": 476000
+    },
+    {
+      "epoch": 21.46,
+      "learning_rate": 1.0579353535353536e-05,
+      "loss": 1.0758,
+      "step": 476500
+    },
+    {
+      "epoch": 21.48,
+      "learning_rate": 1.0569252525252526e-05,
+      "loss": 1.0743,
+      "step": 477000
+    },
+    {
+      "epoch": 21.51,
+      "learning_rate": 1.0559171717171718e-05,
+      "loss": 1.0727,
+      "step": 477500
+    },
+    {
+      "epoch": 21.53,
+      "learning_rate": 1.0549070707070707e-05,
+      "loss": 1.0739,
+      "step": 478000
+    },
+    {
+      "epoch": 21.55,
+      "learning_rate": 1.05389898989899e-05,
+      "loss": 1.0732,
+      "step": 478500
+    },
+    {
+      "epoch": 21.57,
+      "learning_rate": 1.0528888888888891e-05,
+      "loss": 1.0711,
+      "step": 479000
+    },
+    {
+      "epoch": 21.6,
+      "learning_rate": 1.051878787878788e-05,
+      "loss": 1.0733,
+      "step": 479500
+    },
+    {
+      "epoch": 21.62,
+      "learning_rate": 1.050868686868687e-05,
+      "loss": 1.0714,
+      "step": 480000
+    },
+    {
+      "epoch": 21.64,
+      "learning_rate": 1.0498585858585859e-05,
+      "loss": 1.071,
+      "step": 480500
+    },
+    {
+      "epoch": 21.66,
+      "learning_rate": 1.048848484848485e-05,
+      "loss": 1.0738,
+      "step": 481000
+    },
+    {
+      "epoch": 21.69,
+      "learning_rate": 1.047838383838384e-05,
+      "loss": 1.0712,
+      "step": 481500
+    },
+    {
+      "epoch": 21.71,
+      "learning_rate": 1.0468303030303032e-05,
+      "loss": 1.0706,
+      "step": 482000
+    },
+    {
+      "epoch": 21.73,
+      "learning_rate": 1.0458202020202022e-05,
+      "loss": 1.0715,
+      "step": 482500
+    },
+    {
+      "epoch": 21.75,
+      "learning_rate": 1.0448121212121214e-05,
+      "loss": 1.0717,
+      "step": 483000
+    },
+    {
+      "epoch": 21.78,
+      "learning_rate": 1.0438020202020203e-05,
+      "loss": 1.0725,
+      "step": 483500
+    },
+    {
+      "epoch": 21.8,
+      "learning_rate": 1.0427919191919193e-05,
+      "loss": 1.0718,
+      "step": 484000
+    },
+    {
+      "epoch": 21.82,
+      "learning_rate": 1.0417818181818182e-05,
+      "loss": 1.0746,
+      "step": 484500
+    },
+    {
+      "epoch": 21.84,
+      "learning_rate": 1.0407717171717174e-05,
+      "loss": 1.0705,
+      "step": 485000
+    },
+    {
+      "epoch": 21.87,
+      "learning_rate": 1.0397616161616162e-05,
+      "loss": 1.0713,
+      "step": 485500
+    },
+    {
+      "epoch": 21.89,
+      "learning_rate": 1.0387515151515152e-05,
+      "loss": 1.0728,
+      "step": 486000
+    },
+    {
+      "epoch": 21.91,
+      "learning_rate": 1.0377414141414141e-05,
+      "loss": 1.0705,
+      "step": 486500
+    },
+    {
+      "epoch": 21.93,
+      "learning_rate": 1.0367313131313133e-05,
+      "loss": 1.0711,
+      "step": 487000
+    },
+    {
+      "epoch": 21.96,
+      "learning_rate": 1.0357232323232325e-05,
+      "loss": 1.071,
+      "step": 487500
+    },
+    {
+      "epoch": 21.98,
+      "learning_rate": 1.0347131313131314e-05,
+      "loss": 1.0736,
+      "step": 488000
+    },
+    {
+      "epoch": 22.0,
+      "eval_accuracy": 0.7854471162405623,
+      "eval_loss": 0.9062856435775757,
+      "eval_runtime": 54.2791,
+      "eval_samples_per_second": 901.084,
+      "eval_steps_per_second": 14.094,
+      "step": 488444
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 1.0337030303030304e-05,
+      "loss": 1.0741,
+      "step": 488500
+    },
+    {
+      "epoch": 22.03,
+      "learning_rate": 1.0326929292929293e-05,
+      "loss": 1.071,
+      "step": 489000
+    },
+    {
+      "epoch": 22.05,
+      "learning_rate": 1.0316828282828285e-05,
+      "loss": 1.0708,
+      "step": 489500
+    },
+    {
+      "epoch": 22.07,
+      "learning_rate": 1.0306747474747475e-05,
+      "loss": 1.0728,
+      "step": 490000
+    },
+    {
+      "epoch": 22.09,
+      "learning_rate": 1.0296646464646466e-05,
+      "loss": 1.0706,
+      "step": 490500
+    },
+    {
+      "epoch": 22.12,
+      "learning_rate": 1.0286545454545456e-05,
+      "loss": 1.0721,
+      "step": 491000
+    },
+    {
+      "epoch": 22.14,
+      "learning_rate": 1.0276444444444444e-05,
+      "loss": 1.0718,
+      "step": 491500
+    },
+    {
+      "epoch": 22.16,
+      "learning_rate": 1.0266343434343435e-05,
+      "loss": 1.067,
+      "step": 492000
+    },
+    {
+      "epoch": 22.18,
+      "learning_rate": 1.0256262626262627e-05,
+      "loss": 1.0713,
+      "step": 492500
+    },
+    {
+      "epoch": 22.21,
+      "learning_rate": 1.0246161616161616e-05,
+      "loss": 1.0705,
+      "step": 493000
+    },
+    {
+      "epoch": 22.23,
+      "learning_rate": 1.0236060606060607e-05,
+      "loss": 1.072,
+      "step": 493500
+    },
+    {
+      "epoch": 22.25,
+      "learning_rate": 1.0225959595959596e-05,
+      "loss": 1.0715,
+      "step": 494000
+    },
+    {
+      "epoch": 22.27,
+      "learning_rate": 1.0215858585858586e-05,
+      "loss": 1.0699,
+      "step": 494500
+    },
+    {
+      "epoch": 22.3,
+      "learning_rate": 1.020579797979798e-05,
+      "loss": 1.0694,
+      "step": 495000
+    },
+    {
+      "epoch": 22.32,
+      "learning_rate": 1.0195717171717173e-05,
+      "loss": 1.0713,
+      "step": 495500
+    },
+    {
+      "epoch": 22.34,
+      "learning_rate": 1.0185616161616163e-05,
+      "loss": 1.0716,
+      "step": 496000
+    },
+    {
+      "epoch": 22.36,
+      "learning_rate": 1.0175515151515152e-05,
+      "loss": 1.0716,
+      "step": 496500
+    },
+    {
+      "epoch": 22.39,
+      "learning_rate": 1.0165414141414142e-05,
+      "loss": 1.0713,
+      "step": 497000
+    },
+    {
+      "epoch": 22.41,
+      "learning_rate": 1.0155313131313132e-05,
+      "loss": 1.0718,
+      "step": 497500
+    },
+    {
+      "epoch": 22.43,
+      "learning_rate": 1.0145212121212123e-05,
+      "loss": 1.0706,
+      "step": 498000
+    },
+    {
+      "epoch": 22.45,
+      "learning_rate": 1.0135111111111111e-05,
+      "loss": 1.0685,
+      "step": 498500
+    },
+    {
+      "epoch": 22.48,
+      "learning_rate": 1.0125010101010101e-05,
+      "loss": 1.0719,
+      "step": 499000
+    },
+    {
+      "epoch": 22.5,
+      "learning_rate": 1.011490909090909e-05,
+      "loss": 1.0699,
+      "step": 499500
+    },
+    {
+      "epoch": 22.52,
+      "learning_rate": 1.0104808080808082e-05,
+      "loss": 1.067,
+      "step": 500000
+    },
+    {
+      "epoch": 22.54,
+      "learning_rate": 1.0094727272727274e-05,
+      "loss": 1.0712,
+      "step": 500500
+    },
+    {
+      "epoch": 22.57,
+      "learning_rate": 1.0084626262626263e-05,
+      "loss": 1.0688,
+      "step": 501000
+    },
+    {
+      "epoch": 22.59,
+      "learning_rate": 1.0074525252525253e-05,
+      "loss": 1.0696,
+      "step": 501500
+    },
+    {
+      "epoch": 22.61,
+      "learning_rate": 1.0064424242424242e-05,
+      "loss": 1.0709,
+      "step": 502000
+    },
+    {
+      "epoch": 22.63,
+      "learning_rate": 1.0054323232323234e-05,
+      "loss": 1.0703,
+      "step": 502500
+    },
+    {
+      "epoch": 22.66,
+      "learning_rate": 1.0044222222222224e-05,
+      "loss": 1.0684,
+      "step": 503000
+    },
+    {
+      "epoch": 22.68,
+      "learning_rate": 1.0034121212121213e-05,
+      "loss": 1.07,
+      "step": 503500
+    },
+    {
+      "epoch": 22.7,
+      "learning_rate": 1.0024040404040405e-05,
+      "loss": 1.0689,
+      "step": 504000
+    },
+    {
+      "epoch": 22.72,
+      "learning_rate": 1.0013939393939395e-05,
+      "loss": 1.0697,
+      "step": 504500
+    },
+    {
+      "epoch": 22.75,
+      "learning_rate": 1.0003838383838384e-05,
+      "loss": 1.0693,
+      "step": 505000
+    },
+    {
+      "epoch": 22.77,
+      "learning_rate": 9.993737373737374e-06,
+      "loss": 1.0704,
+      "step": 505500
+    },
+    {
+      "epoch": 22.79,
+      "learning_rate": 9.983636363636364e-06,
+      "loss": 1.0657,
+      "step": 506000
+    },
+    {
+      "epoch": 22.81,
+      "learning_rate": 9.973555555555557e-06,
+      "loss": 1.0685,
+      "step": 506500
+    },
+    {
+      "epoch": 22.84,
+      "learning_rate": 9.963454545454547e-06,
+      "loss": 1.0679,
+      "step": 507000
+    },
+    {
+      "epoch": 22.86,
+      "learning_rate": 9.953353535353535e-06,
+      "loss": 1.0673,
+      "step": 507500
+    },
+    {
+      "epoch": 22.88,
+      "learning_rate": 9.943252525252526e-06,
+      "loss": 1.0687,
+      "step": 508000
+    },
+    {
+      "epoch": 22.9,
+      "learning_rate": 9.933151515151516e-06,
+      "loss": 1.07,
+      "step": 508500
+    },
+    {
+      "epoch": 22.93,
+      "learning_rate": 9.923070707070708e-06,
+      "loss": 1.0667,
+      "step": 509000
+    },
+    {
+      "epoch": 22.95,
+      "learning_rate": 9.912969696969697e-06,
+      "loss": 1.0671,
+      "step": 509500
+    },
+    {
+      "epoch": 22.97,
+      "learning_rate": 9.902868686868687e-06,
+      "loss": 1.068,
+      "step": 510000
+    },
+    {
+      "epoch": 22.99,
+      "learning_rate": 9.892767676767677e-06,
+      "loss": 1.0702,
+      "step": 510500
+    },
+    {
+      "epoch": 23.0,
+      "eval_accuracy": 0.7856228909334859,
+      "eval_loss": 0.9029592275619507,
+      "eval_runtime": 53.5131,
+      "eval_samples_per_second": 913.982,
+      "eval_steps_per_second": 14.296,
+      "step": 510646
+    },
+    {
+      "epoch": 23.02,
+      "learning_rate": 9.882666666666668e-06,
+      "loss": 1.0661,
+      "step": 511000
+    },
+    {
+      "epoch": 23.04,
+      "learning_rate": 9.872565656565656e-06,
+      "loss": 1.0678,
+      "step": 511500
+    },
+    {
+      "epoch": 23.06,
+      "learning_rate": 9.862464646464647e-06,
+      "loss": 1.0674,
+      "step": 512000
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 9.852363636363637e-06,
+      "loss": 1.066,
+      "step": 512500
+    },
+    {
+      "epoch": 23.11,
+      "learning_rate": 9.842282828282829e-06,
+      "loss": 1.0671,
+      "step": 513000
+    },
+    {
+      "epoch": 23.13,
+      "learning_rate": 9.832181818181818e-06,
+      "loss": 1.0684,
+      "step": 513500
+    },
+    {
+      "epoch": 23.15,
+      "learning_rate": 9.82208080808081e-06,
+      "loss": 1.0681,
+      "step": 514000
+    },
+    {
+      "epoch": 23.17,
+      "learning_rate": 9.812e-06,
+      "loss": 1.0677,
+      "step": 514500
+    },
+    {
+      "epoch": 23.2,
+      "learning_rate": 9.80189898989899e-06,
+      "loss": 1.0665,
+      "step": 515000
+    },
+    {
+      "epoch": 23.22,
+      "learning_rate": 9.79179797979798e-06,
+      "loss": 1.0684,
+      "step": 515500
+    },
+    {
+      "epoch": 23.24,
+      "learning_rate": 9.781696969696971e-06,
+      "loss": 1.0695,
+      "step": 516000
+    },
+    {
+      "epoch": 23.26,
+      "learning_rate": 9.771595959595961e-06,
+      "loss": 1.0671,
+      "step": 516500
+    },
+    {
+      "epoch": 23.29,
+      "learning_rate": 9.76149494949495e-06,
+      "loss": 1.0673,
+      "step": 517000
+    },
+    {
+      "epoch": 23.31,
+      "learning_rate": 9.75139393939394e-06,
+      "loss": 1.0652,
+      "step": 517500
+    },
+    {
+      "epoch": 23.33,
+      "learning_rate": 9.74129292929293e-06,
+      "loss": 1.0672,
+      "step": 518000
+    },
+    {
+      "epoch": 23.35,
+      "learning_rate": 9.73119191919192e-06,
+      "loss": 1.068,
+      "step": 518500
+    },
+    {
+      "epoch": 23.38,
+      "learning_rate": 9.72109090909091e-06,
+      "loss": 1.0664,
+      "step": 519000
+    },
+    {
+      "epoch": 23.4,
+      "learning_rate": 9.7109898989899e-06,
+      "loss": 1.0662,
+      "step": 519500
+    },
+    {
+      "epoch": 23.42,
+      "learning_rate": 9.70088888888889e-06,
+      "loss": 1.0681,
+      "step": 520000
+    },
+    {
+      "epoch": 23.44,
+      "learning_rate": 9.690808080808082e-06,
+      "loss": 1.069,
+      "step": 520500
+    },
+    {
+      "epoch": 23.47,
+      "learning_rate": 9.680707070707071e-06,
+      "loss": 1.0682,
+      "step": 521000
+    },
+    {
+      "epoch": 23.49,
+      "learning_rate": 9.670606060606061e-06,
+      "loss": 1.0692,
+      "step": 521500
+    },
+    {
+      "epoch": 23.51,
+      "learning_rate": 9.660505050505051e-06,
+      "loss": 1.0651,
+      "step": 522000
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 9.650424242424244e-06,
+      "loss": 1.0668,
+      "step": 522500
+    },
+    {
+      "epoch": 23.56,
+      "learning_rate": 9.640323232323234e-06,
+      "loss": 1.0652,
+      "step": 523000
+    },
+    {
+      "epoch": 23.58,
+      "learning_rate": 9.630222222222223e-06,
+      "loss": 1.0677,
+      "step": 523500
+    },
+    {
+      "epoch": 23.6,
+      "learning_rate": 9.620121212121213e-06,
+      "loss": 1.0656,
+      "step": 524000
+    },
+    {
+      "epoch": 23.62,
+      "learning_rate": 9.610020202020203e-06,
+      "loss": 1.0697,
+      "step": 524500
+    },
+    {
+      "epoch": 23.65,
+      "learning_rate": 9.599919191919192e-06,
+      "loss": 1.0665,
+      "step": 525000
+    },
+    {
+      "epoch": 23.67,
+      "learning_rate": 9.589838383838384e-06,
+      "loss": 1.0673,
+      "step": 525500
+    },
+    {
+      "epoch": 23.69,
+      "learning_rate": 9.579737373737374e-06,
+      "loss": 1.0682,
+      "step": 526000
+    },
+    {
+      "epoch": 23.71,
+      "learning_rate": 9.569656565656567e-06,
+      "loss": 1.0665,
+      "step": 526500
+    },
+    {
+      "epoch": 23.74,
+      "learning_rate": 9.559555555555557e-06,
+      "loss": 1.0673,
+      "step": 527000
+    },
+    {
+      "epoch": 23.76,
+      "learning_rate": 9.549454545454545e-06,
+      "loss": 1.0637,
+      "step": 527500
+    },
+    {
+      "epoch": 23.78,
+      "learning_rate": 9.539353535353536e-06,
+      "loss": 1.0654,
+      "step": 528000
+    },
+    {
+      "epoch": 23.8,
+      "learning_rate": 9.529252525252526e-06,
+      "loss": 1.066,
+      "step": 528500
+    },
+    {
+      "epoch": 23.83,
+      "learning_rate": 9.519151515151516e-06,
+      "loss": 1.0654,
+      "step": 529000
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 9.509050505050505e-06,
+      "loss": 1.065,
+      "step": 529500
+    },
+    {
+      "epoch": 23.87,
+      "learning_rate": 9.498949494949495e-06,
+      "loss": 1.066,
+      "step": 530000
+    },
+    {
+      "epoch": 23.89,
+      "learning_rate": 9.488848484848485e-06,
+      "loss": 1.0649,
+      "step": 530500
+    },
+    {
+      "epoch": 23.92,
+      "learning_rate": 9.478747474747476e-06,
+      "loss": 1.0645,
+      "step": 531000
+    },
+    {
+      "epoch": 23.94,
+      "learning_rate": 9.468646464646466e-06,
+      "loss": 1.0661,
+      "step": 531500
+    },
+    {
+      "epoch": 23.96,
+      "learning_rate": 9.458545454545456e-06,
+      "loss": 1.0634,
+      "step": 532000
+    },
+    {
+      "epoch": 23.98,
+      "learning_rate": 9.448464646464647e-06,
+      "loss": 1.0645,
+      "step": 532500
+    },
+    {
+      "epoch": 24.0,
+      "eval_accuracy": 0.7862003491289279,
+      "eval_loss": 0.8998920321464539,
+      "eval_runtime": 54.3047,
+      "eval_samples_per_second": 900.659,
+      "eval_steps_per_second": 14.087,
+      "step": 532848
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 9.438383838383839e-06,
+      "loss": 1.0649,
+      "step": 533000
+    },
+    {
+      "epoch": 24.03,
+      "learning_rate": 9.42828282828283e-06,
+      "loss": 1.0651,
+      "step": 533500
+    },
+    {
+      "epoch": 24.05,
+      "learning_rate": 9.418181818181818e-06,
+      "loss": 1.0639,
+      "step": 534000
+    },
+    {
+      "epoch": 24.07,
+      "learning_rate": 9.408080808080808e-06,
+      "loss": 1.0668,
+      "step": 534500
+    },
+    {
+      "epoch": 24.1,
+      "learning_rate": 9.397979797979799e-06,
+      "loss": 1.0649,
+      "step": 535000
+    },
+    {
+      "epoch": 24.12,
+      "learning_rate": 9.387878787878789e-06,
+      "loss": 1.0649,
+      "step": 535500
+    },
+    {
+      "epoch": 24.14,
+      "learning_rate": 9.37779797979798e-06,
+      "loss": 1.064,
+      "step": 536000
+    },
+    {
+      "epoch": 24.16,
+      "learning_rate": 9.36769696969697e-06,
+      "loss": 1.0634,
+      "step": 536500
+    },
+    {
+      "epoch": 24.19,
+      "learning_rate": 9.357595959595962e-06,
+      "loss": 1.0641,
+      "step": 537000
+    },
+    {
+      "epoch": 24.21,
+      "learning_rate": 9.34749494949495e-06,
+      "loss": 1.0622,
+      "step": 537500
+    },
+    {
+      "epoch": 24.23,
+      "learning_rate": 9.337414141414142e-06,
+      "loss": 1.0644,
+      "step": 538000
+    },
+    {
+      "epoch": 24.25,
+      "learning_rate": 9.327313131313131e-06,
+      "loss": 1.062,
+      "step": 538500
+    },
+    {
+      "epoch": 24.28,
+      "learning_rate": 9.317232323232323e-06,
+      "loss": 1.0674,
+      "step": 539000
+    },
+    {
+      "epoch": 24.3,
+      "learning_rate": 9.307131313131314e-06,
+      "loss": 1.0638,
+      "step": 539500
+    },
+    {
+      "epoch": 24.32,
+      "learning_rate": 9.297030303030304e-06,
+      "loss": 1.0628,
+      "step": 540000
+    },
+    {
+      "epoch": 24.34,
+      "learning_rate": 9.286929292929294e-06,
+      "loss": 1.0658,
+      "step": 540500
+    },
+    {
+      "epoch": 24.37,
+      "learning_rate": 9.276828282828284e-06,
+      "loss": 1.062,
+      "step": 541000
+    },
+    {
+      "epoch": 24.39,
+      "learning_rate": 9.266727272727275e-06,
+      "loss": 1.0664,
+      "step": 541500
+    },
+    {
+      "epoch": 24.41,
+      "learning_rate": 9.256626262626263e-06,
+      "loss": 1.0627,
+      "step": 542000
+    },
+    {
+      "epoch": 24.43,
+      "learning_rate": 9.246525252525254e-06,
+      "loss": 1.0646,
+      "step": 542500
+    },
+    {
+      "epoch": 24.46,
+      "learning_rate": 9.236424242424244e-06,
+      "loss": 1.063,
+      "step": 543000
+    },
+    {
+      "epoch": 24.48,
+      "learning_rate": 9.226323232323233e-06,
+      "loss": 1.0627,
+      "step": 543500
+    },
+    {
+      "epoch": 24.5,
+      "learning_rate": 9.216242424242425e-06,
+      "loss": 1.0643,
+      "step": 544000
+    },
+    {
+      "epoch": 24.52,
+      "learning_rate": 9.206141414141415e-06,
+      "loss": 1.0643,
+      "step": 544500
+    },
+    {
+      "epoch": 24.55,
+      "learning_rate": 9.196040404040405e-06,
+      "loss": 1.0628,
+      "step": 545000
+    },
+    {
+      "epoch": 24.57,
+      "learning_rate": 9.185939393939396e-06,
+      "loss": 1.0642,
+      "step": 545500
+    },
+    {
+      "epoch": 24.59,
+      "learning_rate": 9.175838383838384e-06,
+      "loss": 1.0647,
+      "step": 546000
+    },
+    {
+      "epoch": 24.61,
+      "learning_rate": 9.165737373737374e-06,
+      "loss": 1.0646,
+      "step": 546500
+    },
+    {
+      "epoch": 24.64,
+      "learning_rate": 9.155636363636365e-06,
+      "loss": 1.0615,
+      "step": 547000
+    },
+    {
+      "epoch": 24.66,
+      "learning_rate": 9.145555555555557e-06,
+      "loss": 1.0628,
+      "step": 547500
+    },
+    {
+      "epoch": 24.68,
+      "learning_rate": 9.135454545454546e-06,
+      "loss": 1.0654,
+      "step": 548000
+    },
+    {
+      "epoch": 24.7,
+      "learning_rate": 9.125353535353536e-06,
+      "loss": 1.0638,
+      "step": 548500
+    },
+    {
+      "epoch": 24.73,
+      "learning_rate": 9.115252525252526e-06,
+      "loss": 1.0612,
+      "step": 549000
+    },
+    {
+      "epoch": 24.75,
+      "learning_rate": 9.105151515151516e-06,
+      "loss": 1.0615,
+      "step": 549500
+    },
+    {
+      "epoch": 24.77,
+      "learning_rate": 9.095070707070707e-06,
+      "loss": 1.0649,
+      "step": 550000
+    },
+    {
+      "epoch": 24.8,
+      "learning_rate": 9.084969696969697e-06,
+      "loss": 1.0638,
+      "step": 550500
+    },
+    {
+      "epoch": 24.82,
+      "learning_rate": 9.074868686868688e-06,
+      "loss": 1.0642,
+      "step": 551000
+    },
+    {
+      "epoch": 24.84,
+      "learning_rate": 9.064767676767678e-06,
+      "loss": 1.0647,
+      "step": 551500
+    },
+    {
+      "epoch": 24.86,
+      "learning_rate": 9.054666666666666e-06,
+      "loss": 1.0623,
+      "step": 552000
+    },
+    {
+      "epoch": 24.89,
+      "learning_rate": 9.044565656565657e-06,
+      "loss": 1.0654,
+      "step": 552500
+    },
+    {
+      "epoch": 24.91,
+      "learning_rate": 9.034464646464647e-06,
+      "loss": 1.0616,
+      "step": 553000
+    },
+    {
+      "epoch": 24.93,
+      "learning_rate": 9.024363636363637e-06,
+      "loss": 1.0651,
+      "step": 553500
+    },
+    {
+      "epoch": 24.95,
+      "learning_rate": 9.01428282828283e-06,
+      "loss": 1.0616,
+      "step": 554000
+    },
+    {
+      "epoch": 24.98,
+      "learning_rate": 9.004181818181818e-06,
+      "loss": 1.0629,
+      "step": 554500
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 8.994080808080808e-06,
+      "loss": 1.0636,
+      "step": 555000
+    },
+    {
+      "epoch": 25.0,
+      "eval_accuracy": 0.7867430936321952,
+      "eval_loss": 0.8976159691810608,
+      "eval_runtime": 54.2015,
+      "eval_samples_per_second": 902.374,
+      "eval_steps_per_second": 14.114,
+      "step": 555050
+    },
+    {
+      "epoch": 25.02,
+      "learning_rate": 8.984e-06,
+      "loss": 1.0611,
+      "step": 555500
+    },
+    {
+      "epoch": 25.04,
+      "learning_rate": 8.973898989898991e-06,
+      "loss": 1.0607,
+      "step": 556000
+    },
+    {
+      "epoch": 25.07,
+      "learning_rate": 8.96379797979798e-06,
+      "loss": 1.0621,
+      "step": 556500
+    },
+    {
+      "epoch": 25.09,
+      "learning_rate": 8.953717171717172e-06,
+      "loss": 1.061,
+      "step": 557000
+    },
+    {
+      "epoch": 25.11,
+      "learning_rate": 8.943616161616162e-06,
+      "loss": 1.0621,
+      "step": 557500
+    },
+    {
+      "epoch": 25.13,
+      "learning_rate": 8.933515151515152e-06,
+      "loss": 1.0611,
+      "step": 558000
+    },
+    {
+      "epoch": 25.16,
+      "learning_rate": 8.923414141414141e-06,
+      "loss": 1.0623,
+      "step": 558500
+    },
+    {
+      "epoch": 25.18,
+      "learning_rate": 8.913313131313131e-06,
+      "loss": 1.061,
+      "step": 559000
+    },
+    {
+      "epoch": 25.2,
+      "learning_rate": 8.903212121212122e-06,
+      "loss": 1.0617,
+      "step": 559500
+    },
+    {
+      "epoch": 25.22,
+      "learning_rate": 8.893111111111112e-06,
+      "loss": 1.0619,
+      "step": 560000
+    },
+    {
+      "epoch": 25.25,
+      "learning_rate": 8.883010101010102e-06,
+      "loss": 1.0607,
+      "step": 560500
+    },
+    {
+      "epoch": 25.27,
+      "learning_rate": 8.872909090909092e-06,
+      "loss": 1.0601,
+      "step": 561000
+    },
+    {
+      "epoch": 25.29,
+      "learning_rate": 8.862808080808081e-06,
+      "loss": 1.0612,
+      "step": 561500
+    },
+    {
+      "epoch": 25.31,
+      "learning_rate": 8.852727272727273e-06,
+      "loss": 1.0633,
+      "step": 562000
+    },
+    {
+      "epoch": 25.34,
+      "learning_rate": 8.842626262626264e-06,
+      "loss": 1.0618,
+      "step": 562500
+    },
+    {
+      "epoch": 25.36,
+      "learning_rate": 8.832525252525254e-06,
+      "loss": 1.0604,
+      "step": 563000
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 8.822424242424244e-06,
+      "loss": 1.0606,
+      "step": 563500
+    },
+    {
+      "epoch": 25.4,
+      "learning_rate": 8.812343434343435e-06,
+      "loss": 1.0606,
+      "step": 564000
+    },
+    {
+      "epoch": 25.43,
+      "learning_rate": 8.802242424242425e-06,
+      "loss": 1.0617,
+      "step": 564500
+    },
+    {
+      "epoch": 25.45,
+      "learning_rate": 8.792141414141415e-06,
+      "loss": 1.0607,
+      "step": 565000
+    },
+    {
+      "epoch": 25.47,
+      "learning_rate": 8.782040404040406e-06,
+      "loss": 1.0589,
+      "step": 565500
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 8.771959595959598e-06,
+      "loss": 1.0626,
+      "step": 566000
+    },
+    {
+      "epoch": 25.52,
+      "learning_rate": 8.761858585858586e-06,
+      "loss": 1.0633,
+      "step": 566500
+    },
+    {
+      "epoch": 25.54,
+      "learning_rate": 8.751757575757577e-06,
+      "loss": 1.0629,
+      "step": 567000
+    },
+    {
+      "epoch": 25.56,
+      "learning_rate": 8.741656565656567e-06,
+      "loss": 1.0619,
+      "step": 567500
+    },
+    {
+      "epoch": 25.58,
+      "learning_rate": 8.73157575757576e-06,
+      "loss": 1.0644,
+      "step": 568000
+    },
+    {
+      "epoch": 25.61,
+      "learning_rate": 8.721474747474748e-06,
+      "loss": 1.0594,
+      "step": 568500
+    },
+    {
+      "epoch": 25.63,
+      "learning_rate": 8.711373737373738e-06,
+      "loss": 1.0619,
+      "step": 569000
+    },
+    {
+      "epoch": 25.65,
+      "learning_rate": 8.701272727272728e-06,
+      "loss": 1.0566,
+      "step": 569500
+    },
+    {
+      "epoch": 25.67,
+      "learning_rate": 8.691171717171719e-06,
+      "loss": 1.0611,
+      "step": 570000
+    },
+    {
+      "epoch": 25.7,
+      "learning_rate": 8.681070707070707e-06,
+      "loss": 1.0603,
+      "step": 570500
+    },
+    {
+      "epoch": 25.72,
+      "learning_rate": 8.670969696969698e-06,
+      "loss": 1.0589,
+      "step": 571000
+    },
+    {
+      "epoch": 25.74,
+      "learning_rate": 8.660868686868688e-06,
+      "loss": 1.0611,
+      "step": 571500
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 8.650767676767678e-06,
+      "loss": 1.0611,
+      "step": 572000
+    },
+    {
+      "epoch": 25.79,
+      "learning_rate": 8.640686868686869e-06,
+      "loss": 1.0577,
+      "step": 572500
+    },
+    {
+      "epoch": 25.81,
+      "learning_rate": 8.630585858585859e-06,
+      "loss": 1.0569,
+      "step": 573000
+    },
+    {
+      "epoch": 25.83,
+      "learning_rate": 8.62048484848485e-06,
+      "loss": 1.0598,
+      "step": 573500
+    },
+    {
+      "epoch": 25.85,
+      "learning_rate": 8.61038383838384e-06,
+      "loss": 1.0618,
+      "step": 574000
+    },
+    {
+      "epoch": 25.88,
+      "learning_rate": 8.600303030303032e-06,
+      "loss": 1.0646,
+      "step": 574500
+    },
+    {
+      "epoch": 25.9,
+      "learning_rate": 8.59020202020202e-06,
+      "loss": 1.0602,
+      "step": 575000
+    },
+    {
+      "epoch": 25.92,
+      "learning_rate": 8.580121212121213e-06,
+      "loss": 1.061,
+      "step": 575500
+    },
+    {
+      "epoch": 25.94,
+      "learning_rate": 8.570020202020203e-06,
+      "loss": 1.0641,
+      "step": 576000
+    },
+    {
+      "epoch": 25.97,
+      "learning_rate": 8.559919191919193e-06,
+      "loss": 1.0582,
+      "step": 576500
+    },
+    {
+      "epoch": 25.99,
+      "learning_rate": 8.549818181818182e-06,
+      "loss": 1.0604,
+      "step": 577000
+    },
+    {
+      "epoch": 26.0,
+      "eval_accuracy": 0.7874246127899406,
+      "eval_loss": 0.8951030969619751,
+      "eval_runtime": 53.6605,
+      "eval_samples_per_second": 911.471,
+      "eval_steps_per_second": 14.256,
+      "step": 577252
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 8.539737373737374e-06,
+      "loss": 1.0597,
+      "step": 577500
+    },
+    {
+      "epoch": 26.03,
+      "learning_rate": 8.529636363636364e-06,
+      "loss": 1.0587,
+      "step": 578000
+    },
+    {
+      "epoch": 26.06,
+      "learning_rate": 8.519535353535355e-06,
+      "loss": 1.059,
+      "step": 578500
+    },
+    {
+      "epoch": 26.08,
+      "learning_rate": 8.509434343434343e-06,
+      "loss": 1.0592,
+      "step": 579000
+    },
+    {
+      "epoch": 26.1,
+      "learning_rate": 8.499353535353535e-06,
+      "loss": 1.06,
+      "step": 579500
+    },
+    {
+      "epoch": 26.12,
+      "learning_rate": 8.489252525252526e-06,
+      "loss": 1.0582,
+      "step": 580000
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 8.479151515151516e-06,
+      "loss": 1.0586,
+      "step": 580500
+    },
+    {
+      "epoch": 26.17,
+      "learning_rate": 8.469050505050506e-06,
+      "loss": 1.0582,
+      "step": 581000
+    },
+    {
+      "epoch": 26.19,
+      "learning_rate": 8.458949494949495e-06,
+      "loss": 1.0601,
+      "step": 581500
+    },
+    {
+      "epoch": 26.21,
+      "learning_rate": 8.448848484848485e-06,
+      "loss": 1.0595,
+      "step": 582000
+    },
+    {
+      "epoch": 26.24,
+      "learning_rate": 8.438747474747475e-06,
+      "loss": 1.0595,
+      "step": 582500
+    },
+    {
+      "epoch": 26.26,
+      "learning_rate": 8.428646464646466e-06,
+      "loss": 1.0589,
+      "step": 583000
+    },
+    {
+      "epoch": 26.28,
+      "learning_rate": 8.418545454545454e-06,
+      "loss": 1.0612,
+      "step": 583500
+    },
+    {
+      "epoch": 26.3,
+      "learning_rate": 8.408464646464647e-06,
+      "loss": 1.0587,
+      "step": 584000
+    },
+    {
+      "epoch": 26.33,
+      "learning_rate": 8.398363636363637e-06,
+      "loss": 1.0597,
+      "step": 584500
+    },
+    {
+      "epoch": 26.35,
+      "learning_rate": 8.388262626262627e-06,
+      "loss": 1.0581,
+      "step": 585000
+    },
+    {
+      "epoch": 26.37,
+      "learning_rate": 8.378161616161616e-06,
+      "loss": 1.0582,
+      "step": 585500
+    },
+    {
+      "epoch": 26.39,
+      "learning_rate": 8.368060606060606e-06,
+      "loss": 1.0588,
+      "step": 586000
+    },
+    {
+      "epoch": 26.42,
+      "learning_rate": 8.357979797979798e-06,
+      "loss": 1.0575,
+      "step": 586500
+    },
+    {
+      "epoch": 26.44,
+      "learning_rate": 8.347878787878789e-06,
+      "loss": 1.0613,
+      "step": 587000
+    },
+    {
+      "epoch": 26.46,
+      "learning_rate": 8.337777777777777e-06,
+      "loss": 1.0565,
+      "step": 587500
+    },
+    {
+      "epoch": 26.48,
+      "learning_rate": 8.327676767676767e-06,
+      "loss": 1.0582,
+      "step": 588000
+    },
+    {
+      "epoch": 26.51,
+      "learning_rate": 8.31757575757576e-06,
+      "loss": 1.0575,
+      "step": 588500
+    },
+    {
+      "epoch": 26.53,
+      "learning_rate": 8.307474747474748e-06,
+      "loss": 1.0605,
+      "step": 589000
+    },
+    {
+      "epoch": 26.55,
+      "learning_rate": 8.297373737373738e-06,
+      "loss": 1.0584,
+      "step": 589500
+    },
+    {
+      "epoch": 26.57,
+      "learning_rate": 8.28729292929293e-06,
+      "loss": 1.0571,
+      "step": 590000
+    },
+    {
+      "epoch": 26.6,
+      "learning_rate": 8.277191919191921e-06,
+      "loss": 1.0552,
+      "step": 590500
+    },
+    {
+      "epoch": 26.62,
+      "learning_rate": 8.26709090909091e-06,
+      "loss": 1.058,
+      "step": 591000
+    },
+    {
+      "epoch": 26.64,
+      "learning_rate": 8.257010101010102e-06,
+      "loss": 1.0582,
+      "step": 591500
+    },
+    {
+      "epoch": 26.66,
+      "learning_rate": 8.246909090909092e-06,
+      "loss": 1.0564,
+      "step": 592000
+    },
+    {
+      "epoch": 26.69,
+      "learning_rate": 8.236808080808082e-06,
+      "loss": 1.059,
+      "step": 592500
+    },
+    {
+      "epoch": 26.71,
+      "learning_rate": 8.226707070707073e-06,
+      "loss": 1.0582,
+      "step": 593000
+    },
+    {
+      "epoch": 26.73,
+      "learning_rate": 8.216606060606061e-06,
+      "loss": 1.0568,
+      "step": 593500
+    },
+    {
+      "epoch": 26.75,
+      "learning_rate": 8.206505050505051e-06,
+      "loss": 1.0565,
+      "step": 594000
+    },
+    {
+      "epoch": 26.78,
+      "learning_rate": 8.196404040404042e-06,
+      "loss": 1.0596,
+      "step": 594500
+    },
+    {
+      "epoch": 26.8,
+      "learning_rate": 8.18630303030303e-06,
+      "loss": 1.0571,
+      "step": 595000
+    },
+    {
+      "epoch": 26.82,
+      "learning_rate": 8.176222222222223e-06,
+      "loss": 1.0575,
+      "step": 595500
+    },
+    {
+      "epoch": 26.84,
+      "learning_rate": 8.166121212121213e-06,
+      "loss": 1.0591,
+      "step": 596000
+    },
+    {
+      "epoch": 26.87,
+      "learning_rate": 8.156020202020203e-06,
+      "loss": 1.0575,
+      "step": 596500
+    },
+    {
+      "epoch": 26.89,
+      "learning_rate": 8.145919191919193e-06,
+      "loss": 1.0569,
+      "step": 597000
+    },
+    {
+      "epoch": 26.91,
+      "learning_rate": 8.135838383838384e-06,
+      "loss": 1.0569,
+      "step": 597500
+    },
+    {
+      "epoch": 26.93,
+      "learning_rate": 8.125737373737374e-06,
+      "loss": 1.0594,
+      "step": 598000
+    },
+    {
+      "epoch": 26.96,
+      "learning_rate": 8.115636363636365e-06,
+      "loss": 1.0589,
+      "step": 598500
+    },
+    {
+      "epoch": 26.98,
+      "learning_rate": 8.105535353535355e-06,
+      "loss": 1.0575,
+      "step": 599000
+    },
+    {
+      "epoch": 27.0,
+      "eval_accuracy": 0.7878362218687557,
+      "eval_loss": 0.8923379778862,
+      "eval_runtime": 53.7372,
+      "eval_samples_per_second": 910.17,
+      "eval_steps_per_second": 14.236,
+      "step": 599454
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 8.095454545454547e-06,
+      "loss": 1.0587,
+      "step": 599500
+    },
+    {
+      "epoch": 27.02,
+      "learning_rate": 8.085353535353536e-06,
+      "loss": 1.0598,
+      "step": 600000
+    },
+    {
+      "epoch": 27.05,
+      "learning_rate": 8.075252525252526e-06,
+      "loss": 1.0591,
+      "step": 600500
+    },
+    {
+      "epoch": 27.07,
+      "learning_rate": 8.065151515151516e-06,
+      "loss": 1.0577,
+      "step": 601000
+    },
+    {
+      "epoch": 27.09,
+      "learning_rate": 8.055050505050505e-06,
+      "loss": 1.0562,
+      "step": 601500
+    },
+    {
+      "epoch": 27.11,
+      "learning_rate": 8.044969696969697e-06,
+      "loss": 1.0566,
+      "step": 602000
+    },
+    {
+      "epoch": 27.14,
+      "learning_rate": 8.034868686868687e-06,
+      "loss": 1.0584,
+      "step": 602500
+    },
+    {
+      "epoch": 27.16,
+      "learning_rate": 8.024767676767678e-06,
+      "loss": 1.0565,
+      "step": 603000
+    },
+    {
+      "epoch": 27.18,
+      "learning_rate": 8.014666666666668e-06,
+      "loss": 1.057,
+      "step": 603500
+    },
+    {
+      "epoch": 27.2,
+      "learning_rate": 8.004565656565657e-06,
+      "loss": 1.0578,
+      "step": 604000
+    },
+    {
+      "epoch": 27.23,
+      "learning_rate": 7.994484848484849e-06,
+      "loss": 1.0565,
+      "step": 604500
+    },
+    {
+      "epoch": 27.25,
+      "learning_rate": 7.984383838383839e-06,
+      "loss": 1.0565,
+      "step": 605000
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 7.97428282828283e-06,
+      "loss": 1.0566,
+      "step": 605500
+    },
+    {
+      "epoch": 27.29,
+      "learning_rate": 7.964181818181818e-06,
+      "loss": 1.0571,
+      "step": 606000
+    },
+    {
+      "epoch": 27.32,
+      "learning_rate": 7.954080808080808e-06,
+      "loss": 1.0543,
+      "step": 606500
+    },
+    {
+      "epoch": 27.34,
+      "learning_rate": 7.944e-06,
+      "loss": 1.0571,
+      "step": 607000
+    },
+    {
+      "epoch": 27.36,
+      "learning_rate": 7.93389898989899e-06,
+      "loss": 1.0551,
+      "step": 607500
+    },
+    {
+      "epoch": 27.38,
+      "learning_rate": 7.923797979797981e-06,
+      "loss": 1.0578,
+      "step": 608000
+    },
+    {
+      "epoch": 27.41,
+      "learning_rate": 7.91369696969697e-06,
+      "loss": 1.0546,
+      "step": 608500
+    },
+    {
+      "epoch": 27.43,
+      "learning_rate": 7.903616161616162e-06,
+      "loss": 1.0554,
+      "step": 609000
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 7.893515151515152e-06,
+      "loss": 1.0592,
+      "step": 609500
+    },
+    {
+      "epoch": 27.48,
+      "learning_rate": 7.883414141414143e-06,
+      "loss": 1.0574,
+      "step": 610000
+    },
+    {
+      "epoch": 27.5,
+      "learning_rate": 7.873313131313131e-06,
+      "loss": 1.0549,
+      "step": 610500
+    },
+    {
+      "epoch": 27.52,
+      "learning_rate": 7.863212121212121e-06,
+      "loss": 1.0562,
+      "step": 611000
+    },
+    {
+      "epoch": 27.54,
+      "learning_rate": 7.853111111111112e-06,
+      "loss": 1.0551,
+      "step": 611500
+    },
+    {
+      "epoch": 27.57,
+      "learning_rate": 7.843010101010102e-06,
+      "loss": 1.0541,
+      "step": 612000
+    },
+    {
+      "epoch": 27.59,
+      "learning_rate": 7.83290909090909e-06,
+      "loss": 1.0559,
+      "step": 612500
+    },
+    {
+      "epoch": 27.61,
+      "learning_rate": 7.822808080808083e-06,
+      "loss": 1.0567,
+      "step": 613000
+    },
+    {
+      "epoch": 27.63,
+      "learning_rate": 7.812707070707071e-06,
+      "loss": 1.0549,
+      "step": 613500
+    },
+    {
+      "epoch": 27.66,
+      "learning_rate": 7.802626262626263e-06,
+      "loss": 1.0549,
+      "step": 614000
+    },
+    {
+      "epoch": 27.68,
+      "learning_rate": 7.792525252525254e-06,
+      "loss": 1.0554,
+      "step": 614500
+    },
+    {
+      "epoch": 27.7,
+      "learning_rate": 7.782424242424244e-06,
+      "loss": 1.0562,
+      "step": 615000
+    },
+    {
+      "epoch": 27.72,
+      "learning_rate": 7.772323232323234e-06,
+      "loss": 1.0545,
+      "step": 615500
+    },
+    {
+      "epoch": 27.75,
+      "learning_rate": 7.762222222222223e-06,
+      "loss": 1.058,
+      "step": 616000
+    },
+    {
+      "epoch": 27.77,
+      "learning_rate": 7.752121212121213e-06,
+      "loss": 1.0539,
+      "step": 616500
+    },
+    {
+      "epoch": 27.79,
+      "learning_rate": 7.742020202020203e-06,
+      "loss": 1.0571,
+      "step": 617000
+    },
+    {
+      "epoch": 27.81,
+      "learning_rate": 7.731939393939396e-06,
+      "loss": 1.0553,
+      "step": 617500
+    },
+    {
+      "epoch": 27.84,
+      "learning_rate": 7.721838383838384e-06,
+      "loss": 1.0567,
+      "step": 618000
+    },
+    {
+      "epoch": 27.86,
+      "learning_rate": 7.711737373737375e-06,
+      "loss": 1.0559,
+      "step": 618500
+    },
+    {
+      "epoch": 27.88,
+      "learning_rate": 7.701636363636365e-06,
+      "loss": 1.0556,
+      "step": 619000
+    },
+    {
+      "epoch": 27.9,
+      "learning_rate": 7.691535353535355e-06,
+      "loss": 1.0583,
+      "step": 619500
+    },
+    {
+      "epoch": 27.93,
+      "learning_rate": 7.681434343434344e-06,
+      "loss": 1.0555,
+      "step": 620000
+    },
+    {
+      "epoch": 27.95,
+      "learning_rate": 7.671333333333334e-06,
+      "loss": 1.0554,
+      "step": 620500
+    },
+    {
+      "epoch": 27.97,
+      "learning_rate": 7.661232323232324e-06,
+      "loss": 1.0535,
+      "step": 621000
+    },
+    {
+      "epoch": 27.99,
+      "learning_rate": 7.651151515151517e-06,
+      "loss": 1.0553,
+      "step": 621500
+    },
+    {
+      "epoch": 28.0,
+      "eval_accuracy": 0.7876163608391318,
+      "eval_loss": 0.8919406533241272,
+      "eval_runtime": 58.5229,
+      "eval_samples_per_second": 835.741,
+      "eval_steps_per_second": 13.072,
+      "step": 621656
+    },
+    {
+      "epoch": 28.02,
+      "learning_rate": 7.641050505050505e-06,
+      "loss": 1.0556,
+      "step": 622000
+    },
+    {
+      "epoch": 28.04,
+      "learning_rate": 7.630949494949495e-06,
+      "loss": 1.0539,
+      "step": 622500
+    },
+    {
+      "epoch": 28.06,
+      "learning_rate": 7.620848484848486e-06,
+      "loss": 1.0529,
+      "step": 623000
+    },
+    {
+      "epoch": 28.08,
+      "learning_rate": 7.610747474747475e-06,
+      "loss": 1.0553,
+      "step": 623500
+    },
+    {
+      "epoch": 28.11,
+      "learning_rate": 7.600666666666667e-06,
+      "loss": 1.0562,
+      "step": 624000
+    },
+    {
+      "epoch": 28.13,
+      "learning_rate": 7.590565656565657e-06,
+      "loss": 1.0553,
+      "step": 624500
+    },
+    {
+      "epoch": 28.15,
+      "learning_rate": 7.580464646464647e-06,
+      "loss": 1.0539,
+      "step": 625000
+    },
+    {
+      "epoch": 28.17,
+      "learning_rate": 7.5703636363636365e-06,
+      "loss": 1.0535,
+      "step": 625500
+    },
+    {
+      "epoch": 28.2,
+      "learning_rate": 7.560282828282829e-06,
+      "loss": 1.0538,
+      "step": 626000
+    },
+    {
+      "epoch": 28.22,
+      "learning_rate": 7.550181818181819e-06,
+      "loss": 1.0528,
+      "step": 626500
+    },
+    {
+      "epoch": 28.24,
+      "learning_rate": 7.5400808080808085e-06,
+      "loss": 1.0541,
+      "step": 627000
+    },
+    {
+      "epoch": 28.26,
+      "learning_rate": 7.529979797979799e-06,
+      "loss": 1.0538,
+      "step": 627500
+    },
+    {
+      "epoch": 28.29,
+      "learning_rate": 7.51989898989899e-06,
+      "loss": 1.0523,
+      "step": 628000
+    },
+    {
+      "epoch": 28.31,
+      "learning_rate": 7.5097979797979805e-06,
+      "loss": 1.0556,
+      "step": 628500
+    },
+    {
+      "epoch": 28.33,
+      "learning_rate": 7.49969696969697e-06,
+      "loss": 1.0553,
+      "step": 629000
+    },
+    {
+      "epoch": 28.35,
+      "learning_rate": 7.48959595959596e-06,
+      "loss": 1.0542,
+      "step": 629500
+    },
+    {
+      "epoch": 28.38,
+      "learning_rate": 7.47949494949495e-06,
+      "loss": 1.0558,
+      "step": 630000
+    },
+    {
+      "epoch": 28.4,
+      "learning_rate": 7.46939393939394e-06,
+      "loss": 1.0539,
+      "step": 630500
+    },
+    {
+      "epoch": 28.42,
+      "learning_rate": 7.459292929292929e-06,
+      "loss": 1.0546,
+      "step": 631000
+    },
+    {
+      "epoch": 28.44,
+      "learning_rate": 7.44919191919192e-06,
+      "loss": 1.0545,
+      "step": 631500
+    },
+    {
+      "epoch": 28.47,
+      "learning_rate": 7.439111111111111e-06,
+      "loss": 1.056,
+      "step": 632000
+    },
+    {
+      "epoch": 28.49,
+      "learning_rate": 7.429010101010101e-06,
+      "loss": 1.0521,
+      "step": 632500
+    },
+    {
+      "epoch": 28.51,
+      "learning_rate": 7.418909090909091e-06,
+      "loss": 1.0541,
+      "step": 633000
+    },
+    {
+      "epoch": 28.53,
+      "learning_rate": 7.408828282828283e-06,
+      "loss": 1.055,
+      "step": 633500
+    },
+    {
+      "epoch": 28.56,
+      "learning_rate": 7.398747474747475e-06,
+      "loss": 1.057,
+      "step": 634000
+    },
+    {
+      "epoch": 28.58,
+      "learning_rate": 7.388646464646465e-06,
+      "loss": 1.053,
+      "step": 634500
+    },
+    {
+      "epoch": 28.6,
+      "learning_rate": 7.378545454545455e-06,
+      "loss": 1.0516,
+      "step": 635000
+    },
+    {
+      "epoch": 28.62,
+      "learning_rate": 7.3684444444444444e-06,
+      "loss": 1.0511,
+      "step": 635500
+    },
+    {
+      "epoch": 28.65,
+      "learning_rate": 7.358343434343435e-06,
+      "loss": 1.0552,
+      "step": 636000
+    },
+    {
+      "epoch": 28.67,
+      "learning_rate": 7.348242424242424e-06,
+      "loss": 1.0556,
+      "step": 636500
+    },
+    {
+      "epoch": 28.69,
+      "learning_rate": 7.3381414141414145e-06,
+      "loss": 1.0545,
+      "step": 637000
+    },
+    {
+      "epoch": 28.71,
+      "learning_rate": 7.328040404040406e-06,
+      "loss": 1.053,
+      "step": 637500
+    },
+    {
+      "epoch": 28.74,
+      "learning_rate": 7.317939393939395e-06,
+      "loss": 1.053,
+      "step": 638000
+    },
+    {
+      "epoch": 28.76,
+      "learning_rate": 7.3078383838383845e-06,
+      "loss": 1.0543,
+      "step": 638500
+    },
+    {
+      "epoch": 28.78,
+      "learning_rate": 7.297757575757576e-06,
+      "loss": 1.0542,
+      "step": 639000
+    },
+    {
+      "epoch": 28.8,
+      "learning_rate": 7.287656565656567e-06,
+      "loss": 1.0519,
+      "step": 639500
+    },
+    {
+      "epoch": 28.83,
+      "learning_rate": 7.2775555555555564e-06,
+      "loss": 1.0527,
+      "step": 640000
+    },
+    {
+      "epoch": 28.85,
+      "learning_rate": 7.267454545454547e-06,
+      "loss": 1.0536,
+      "step": 640500
+    },
+    {
+      "epoch": 28.87,
+      "learning_rate": 7.257353535353536e-06,
+      "loss": 1.0532,
+      "step": 641000
+    },
+    {
+      "epoch": 28.89,
+      "learning_rate": 7.2472525252525265e-06,
+      "loss": 1.0535,
+      "step": 641500
+    },
+    {
+      "epoch": 28.92,
+      "learning_rate": 7.237151515151516e-06,
+      "loss": 1.055,
+      "step": 642000
+    },
+    {
+      "epoch": 28.94,
+      "learning_rate": 7.227070707070708e-06,
+      "loss": 1.0536,
+      "step": 642500
+    },
+    {
+      "epoch": 28.96,
+      "learning_rate": 7.216969696969698e-06,
+      "loss": 1.0549,
+      "step": 643000
+    },
+    {
+      "epoch": 28.98,
+      "learning_rate": 7.206868686868688e-06,
+      "loss": 1.0539,
+      "step": 643500
+    },
+    {
+      "epoch": 29.0,
+      "eval_accuracy": 0.787582442453086,
+      "eval_loss": 0.8912253379821777,
+      "eval_runtime": 58.9068,
+      "eval_samples_per_second": 830.295,
+      "eval_steps_per_second": 12.987,
+      "step": 643858
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 7.196767676767677e-06,
+      "loss": 1.0552,
+      "step": 644000
+    },
+    {
+      "epoch": 29.03,
+      "learning_rate": 7.186666666666668e-06,
+      "loss": 1.0528,
+      "step": 644500
+    },
+    {
+      "epoch": 29.05,
+      "learning_rate": 7.176565656565657e-06,
+      "loss": 1.0533,
+      "step": 645000
+    },
+    {
+      "epoch": 29.07,
+      "learning_rate": 7.166464646464647e-06,
+      "loss": 1.0527,
+      "step": 645500
+    },
+    {
+      "epoch": 29.1,
+      "learning_rate": 7.156363636363637e-06,
+      "loss": 1.0513,
+      "step": 646000
+    },
+    {
+      "epoch": 29.12,
+      "learning_rate": 7.146282828282829e-06,
+      "loss": 1.053,
+      "step": 646500
+    },
+    {
+      "epoch": 29.14,
+      "learning_rate": 7.1361818181818185e-06,
+      "loss": 1.0499,
+      "step": 647000
+    },
+    {
+      "epoch": 29.16,
+      "learning_rate": 7.126101010101011e-06,
+      "loss": 1.0527,
+      "step": 647500
+    },
+    {
+      "epoch": 29.19,
+      "learning_rate": 7.116000000000001e-06,
+      "loss": 1.0538,
+      "step": 648000
+    },
+    {
+      "epoch": 29.21,
+      "learning_rate": 7.1058989898989904e-06,
+      "loss": 1.0494,
+      "step": 648500
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 7.095797979797981e-06,
+      "loss": 1.0529,
+      "step": 649000
+    },
+    {
+      "epoch": 29.25,
+      "learning_rate": 7.08569696969697e-06,
+      "loss": 1.0522,
+      "step": 649500
+    },
+    {
+      "epoch": 29.28,
+      "learning_rate": 7.0755959595959604e-06,
+      "loss": 1.0525,
+      "step": 650000
+    },
+    {
+      "epoch": 29.3,
+      "learning_rate": 7.06549494949495e-06,
+      "loss": 1.0514,
+      "step": 650500
+    },
+    {
+      "epoch": 29.32,
+      "learning_rate": 7.055393939393939e-06,
+      "loss": 1.0514,
+      "step": 651000
+    },
+    {
+      "epoch": 29.34,
+      "learning_rate": 7.0453131313131316e-06,
+      "loss": 1.0494,
+      "step": 651500
+    },
+    {
+      "epoch": 29.37,
+      "learning_rate": 7.035232323232324e-06,
+      "loss": 1.0506,
+      "step": 652000
+    },
+    {
+      "epoch": 29.39,
+      "learning_rate": 7.025151515151516e-06,
+      "loss": 1.0515,
+      "step": 652500
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 7.0150505050505055e-06,
+      "loss": 1.051,
+      "step": 653000
+    },
+    {
+      "epoch": 29.43,
+      "learning_rate": 7.004949494949496e-06,
+      "loss": 1.0533,
+      "step": 653500
+    },
+    {
+      "epoch": 29.46,
+      "learning_rate": 6.994848484848485e-06,
+      "loss": 1.0517,
+      "step": 654000
+    },
+    {
+      "epoch": 29.48,
+      "learning_rate": 6.9847474747474755e-06,
+      "loss": 1.0491,
+      "step": 654500
+    },
+    {
+      "epoch": 29.5,
+      "learning_rate": 6.974646464646465e-06,
+      "loss": 1.0509,
+      "step": 655000
+    },
+    {
+      "epoch": 29.52,
+      "learning_rate": 6.964545454545455e-06,
+      "loss": 1.0508,
+      "step": 655500
+    },
+    {
+      "epoch": 29.55,
+      "learning_rate": 6.954444444444445e-06,
+      "loss": 1.0549,
+      "step": 656000
+    },
+    {
+      "epoch": 29.57,
+      "learning_rate": 6.944343434343435e-06,
+      "loss": 1.0494,
+      "step": 656500
+    },
+    {
+      "epoch": 29.59,
+      "learning_rate": 6.934262626262626e-06,
+      "loss": 1.0521,
+      "step": 657000
+    },
+    {
+      "epoch": 29.61,
+      "learning_rate": 6.924161616161617e-06,
+      "loss": 1.0527,
+      "step": 657500
+    },
+    {
+      "epoch": 29.64,
+      "learning_rate": 6.914060606060606e-06,
+      "loss": 1.0511,
+      "step": 658000
+    },
+    {
+      "epoch": 29.66,
+      "learning_rate": 6.903959595959596e-06,
+      "loss": 1.054,
+      "step": 658500
+    },
+    {
+      "epoch": 29.68,
+      "learning_rate": 6.893858585858586e-06,
+      "loss": 1.0522,
+      "step": 659000
+    },
+    {
+      "epoch": 29.7,
+      "learning_rate": 6.883757575757576e-06,
+      "loss": 1.0513,
+      "step": 659500
+    },
+    {
+      "epoch": 29.73,
+      "learning_rate": 6.8736565656565655e-06,
+      "loss": 1.0515,
+      "step": 660000
+    },
+    {
+      "epoch": 29.75,
+      "learning_rate": 6.863555555555557e-06,
+      "loss": 1.0498,
+      "step": 660500
+    },
+    {
+      "epoch": 29.77,
+      "learning_rate": 6.853474747474747e-06,
+      "loss": 1.0522,
+      "step": 661000
+    },
+    {
+      "epoch": 29.79,
+      "learning_rate": 6.8433737373737375e-06,
+      "loss": 1.0523,
+      "step": 661500
+    },
+    {
+      "epoch": 29.82,
+      "learning_rate": 6.833272727272727e-06,
+      "loss": 1.05,
+      "step": 662000
+    },
+    {
+      "epoch": 29.84,
+      "learning_rate": 6.823191919191919e-06,
+      "loss": 1.0511,
+      "step": 662500
+    },
+    {
+      "epoch": 29.86,
+      "learning_rate": 6.8130909090909095e-06,
+      "loss": 1.0556,
+      "step": 663000
+    },
+    {
+      "epoch": 29.88,
+      "learning_rate": 6.802989898989899e-06,
+      "loss": 1.0505,
+      "step": 663500
+    },
+    {
+      "epoch": 29.91,
+      "learning_rate": 6.79288888888889e-06,
+      "loss": 1.053,
+      "step": 664000
+    },
+    {
+      "epoch": 29.93,
+      "learning_rate": 6.7827878787878795e-06,
+      "loss": 1.0523,
+      "step": 664500
+    },
+    {
+      "epoch": 29.95,
+      "learning_rate": 6.77268686868687e-06,
+      "loss": 1.0524,
+      "step": 665000
+    },
+    {
+      "epoch": 29.97,
+      "learning_rate": 6.762585858585859e-06,
+      "loss": 1.0526,
+      "step": 665500
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 6.7524848484848495e-06,
+      "loss": 1.0503,
+      "step": 666000
+    },
+    {
+      "epoch": 30.0,
+      "eval_accuracy": 0.7876511726111519,
+      "eval_loss": 0.8922644853591919,
+      "eval_runtime": 58.5722,
+      "eval_samples_per_second": 835.037,
+      "eval_steps_per_second": 13.061,
+      "step": 666060
+    },
+    {
+      "epoch": 30.02,
+      "learning_rate": 6.742383838383839e-06,
+      "loss": 1.0489,
+      "step": 666500
+    },
+    {
+      "epoch": 30.04,
+      "learning_rate": 6.732303030303031e-06,
+      "loss": 1.0506,
+      "step": 667000
+    },
+    {
+      "epoch": 30.06,
+      "learning_rate": 6.722202020202021e-06,
+      "loss": 1.0511,
+      "step": 667500
+    },
+    {
+      "epoch": 30.09,
+      "learning_rate": 6.712121212121213e-06,
+      "loss": 1.0498,
+      "step": 668000
+    },
+    {
+      "epoch": 30.11,
+      "learning_rate": 6.702020202020203e-06,
+      "loss": 1.0499,
+      "step": 668500
+    },
+    {
+      "epoch": 30.13,
+      "learning_rate": 6.691919191919193e-06,
+      "loss": 1.0482,
+      "step": 669000
+    },
+    {
+      "epoch": 30.15,
+      "learning_rate": 6.681818181818183e-06,
+      "loss": 1.0494,
+      "step": 669500
+    },
+    {
+      "epoch": 30.18,
+      "learning_rate": 6.671717171717172e-06,
+      "loss": 1.0493,
+      "step": 670000
+    },
+    {
+      "epoch": 30.2,
+      "learning_rate": 6.661636363636365e-06,
+      "loss": 1.0507,
+      "step": 670500
+    },
+    {
+      "epoch": 30.22,
+      "learning_rate": 6.651535353535354e-06,
+      "loss": 1.0488,
+      "step": 671000
+    },
+    {
+      "epoch": 30.25,
+      "learning_rate": 6.641434343434344e-06,
+      "loss": 1.0499,
+      "step": 671500
+    },
+    {
+      "epoch": 30.27,
+      "learning_rate": 6.631333333333334e-06,
+      "loss": 1.0493,
+      "step": 672000
+    },
+    {
+      "epoch": 30.29,
+      "learning_rate": 6.621252525252526e-06,
+      "loss": 1.0487,
+      "step": 672500
+    },
+    {
+      "epoch": 30.31,
+      "learning_rate": 6.611151515151516e-06,
+      "loss": 1.0518,
+      "step": 673000
+    },
+    {
+      "epoch": 30.34,
+      "learning_rate": 6.601050505050506e-06,
+      "loss": 1.0522,
+      "step": 673500
+    },
+    {
+      "epoch": 30.36,
+      "learning_rate": 6.590949494949496e-06,
+      "loss": 1.0493,
+      "step": 674000
+    },
+    {
+      "epoch": 30.38,
+      "learning_rate": 6.5808484848484855e-06,
+      "loss": 1.0527,
+      "step": 674500
+    },
+    {
+      "epoch": 30.4,
+      "learning_rate": 6.570747474747475e-06,
+      "loss": 1.0512,
+      "step": 675000
+    },
+    {
+      "epoch": 30.43,
+      "learning_rate": 6.560646464646465e-06,
+      "loss": 1.0511,
+      "step": 675500
+    },
+    {
+      "epoch": 30.45,
+      "learning_rate": 6.550545454545455e-06,
+      "loss": 1.0466,
+      "step": 676000
+    },
+    {
+      "epoch": 30.47,
+      "learning_rate": 6.540444444444445e-06,
+      "loss": 1.0482,
+      "step": 676500
+    },
+    {
+      "epoch": 30.49,
+      "learning_rate": 6.530343434343434e-06,
+      "loss": 1.0497,
+      "step": 677000
+    },
+    {
+      "epoch": 30.52,
+      "learning_rate": 6.520262626262627e-06,
+      "loss": 1.0493,
+      "step": 677500
+    },
+    {
+      "epoch": 30.54,
+      "learning_rate": 6.510161616161617e-06,
+      "loss": 1.0488,
+      "step": 678000
+    },
+    {
+      "epoch": 30.56,
+      "learning_rate": 6.500060606060606e-06,
+      "loss": 1.049,
+      "step": 678500
+    },
+    {
+      "epoch": 30.58,
+      "learning_rate": 6.489959595959597e-06,
+      "loss": 1.0507,
+      "step": 679000
+    },
+    {
+      "epoch": 30.61,
+      "learning_rate": 6.479858585858586e-06,
+      "loss": 1.0477,
+      "step": 679500
+    },
+    {
+      "epoch": 30.63,
+      "learning_rate": 6.4697575757575755e-06,
+      "loss": 1.0504,
+      "step": 680000
+    },
+    {
+      "epoch": 30.65,
+      "learning_rate": 6.459656565656566e-06,
+      "loss": 1.0481,
+      "step": 680500
+    },
+    {
+      "epoch": 30.67,
+      "learning_rate": 6.449575757575758e-06,
+      "loss": 1.0502,
+      "step": 681000
+    },
+    {
+      "epoch": 30.7,
+      "learning_rate": 6.4394747474747475e-06,
+      "loss": 1.0495,
+      "step": 681500
+    },
+    {
+      "epoch": 30.72,
+      "learning_rate": 6.429373737373738e-06,
+      "loss": 1.0482,
+      "step": 682000
+    },
+    {
+      "epoch": 30.74,
+      "learning_rate": 6.419272727272727e-06,
+      "loss": 1.0498,
+      "step": 682500
+    },
+    {
+      "epoch": 30.76,
+      "learning_rate": 6.4091717171717175e-06,
+      "loss": 1.0533,
+      "step": 683000
+    },
+    {
+      "epoch": 30.79,
+      "learning_rate": 6.399090909090909e-06,
+      "loss": 1.047,
+      "step": 683500
+    },
+    {
+      "epoch": 30.81,
+      "learning_rate": 6.388989898989899e-06,
+      "loss": 1.0488,
+      "step": 684000
+    },
+    {
+      "epoch": 30.83,
+      "learning_rate": 6.378888888888889e-06,
+      "loss": 1.051,
+      "step": 684500
+    },
+    {
+      "epoch": 30.85,
+      "learning_rate": 6.36878787878788e-06,
+      "loss": 1.0488,
+      "step": 685000
+    },
+    {
+      "epoch": 30.88,
+      "learning_rate": 6.35868686868687e-06,
+      "loss": 1.0501,
+      "step": 685500
+    },
+    {
+      "epoch": 30.9,
+      "learning_rate": 6.3485858585858595e-06,
+      "loss": 1.0521,
+      "step": 686000
+    },
+    {
+      "epoch": 30.92,
+      "learning_rate": 6.338505050505051e-06,
+      "loss": 1.0502,
+      "step": 686500
+    },
+    {
+      "epoch": 30.94,
+      "learning_rate": 6.328404040404041e-06,
+      "loss": 1.0487,
+      "step": 687000
+    },
+    {
+      "epoch": 30.97,
+      "learning_rate": 6.3183030303030314e-06,
+      "loss": 1.046,
+      "step": 687500
+    },
+    {
+      "epoch": 30.99,
+      "learning_rate": 6.308202020202021e-06,
+      "loss": 1.0465,
+      "step": 688000
+    },
+    {
+      "epoch": 31.0,
+      "eval_accuracy": 0.7886704795527908,
+      "eval_loss": 0.8870713710784912,
+      "eval_runtime": 59.2197,
+      "eval_samples_per_second": 825.907,
+      "eval_steps_per_second": 12.918,
+      "step": 688262
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 6.298121212121213e-06,
+      "loss": 1.0481,
+      "step": 688500
+    },
+    {
+      "epoch": 31.03,
+      "learning_rate": 6.288020202020203e-06,
+      "loss": 1.0494,
+      "step": 689000
+    },
+    {
+      "epoch": 31.06,
+      "learning_rate": 6.277939393939394e-06,
+      "loss": 1.0466,
+      "step": 689500
+    },
+    {
+      "epoch": 31.08,
+      "learning_rate": 6.267838383838383e-06,
+      "loss": 1.049,
+      "step": 690000
+    },
+    {
+      "epoch": 31.1,
+      "learning_rate": 6.2577373737373745e-06,
+      "loss": 1.0474,
+      "step": 690500
+    },
+    {
+      "epoch": 31.12,
+      "learning_rate": 6.247636363636365e-06,
+      "loss": 1.0485,
+      "step": 691000
+    },
+    {
+      "epoch": 31.15,
+      "learning_rate": 6.237535353535354e-06,
+      "loss": 1.0479,
+      "step": 691500
+    },
+    {
+      "epoch": 31.17,
+      "learning_rate": 6.2274343434343446e-06,
+      "loss": 1.0495,
+      "step": 692000
+    },
+    {
+      "epoch": 31.19,
+      "learning_rate": 6.217333333333334e-06,
+      "loss": 1.047,
+      "step": 692500
+    },
+    {
+      "epoch": 31.21,
+      "learning_rate": 6.207232323232324e-06,
+      "loss": 1.0478,
+      "step": 693000
+    },
+    {
+      "epoch": 31.24,
+      "learning_rate": 6.197131313131314e-06,
+      "loss": 1.047,
+      "step": 693500
+    },
+    {
+      "epoch": 31.26,
+      "learning_rate": 6.187030303030304e-06,
+      "loss": 1.0506,
+      "step": 694000
+    },
+    {
+      "epoch": 31.28,
+      "learning_rate": 6.176969696969698e-06,
+      "loss": 1.0471,
+      "step": 694500
+    },
+    {
+      "epoch": 31.3,
+      "learning_rate": 6.166868686868688e-06,
+      "loss": 1.0483,
+      "step": 695000
+    },
+    {
+      "epoch": 31.33,
+      "learning_rate": 6.156767676767678e-06,
+      "loss": 1.0472,
+      "step": 695500
+    },
+    {
+      "epoch": 31.35,
+      "learning_rate": 6.146666666666667e-06,
+      "loss": 1.0511,
+      "step": 696000
+    },
+    {
+      "epoch": 31.37,
+      "learning_rate": 6.136565656565658e-06,
+      "loss": 1.0502,
+      "step": 696500
+    },
+    {
+      "epoch": 31.39,
+      "learning_rate": 6.126464646464647e-06,
+      "loss": 1.0479,
+      "step": 697000
+    },
+    {
+      "epoch": 31.42,
+      "learning_rate": 6.1163636363636366e-06,
+      "loss": 1.0484,
+      "step": 697500
+    },
+    {
+      "epoch": 31.44,
+      "learning_rate": 6.106262626262627e-06,
+      "loss": 1.0465,
+      "step": 698000
+    },
+    {
+      "epoch": 31.46,
+      "learning_rate": 6.096161616161616e-06,
+      "loss": 1.0472,
+      "step": 698500
+    },
+    {
+      "epoch": 31.48,
+      "learning_rate": 6.0860606060606066e-06,
+      "loss": 1.0471,
+      "step": 699000
+    },
+    {
+      "epoch": 31.51,
+      "learning_rate": 6.075959595959596e-06,
+      "loss": 1.05,
+      "step": 699500
+    },
+    {
+      "epoch": 31.53,
+      "learning_rate": 6.065878787878788e-06,
+      "loss": 1.0488,
+      "step": 700000
+    },
+    {
+      "epoch": 31.55,
+      "learning_rate": 6.0557777777777785e-06,
+      "loss": 1.0489,
+      "step": 700500
+    },
+    {
+      "epoch": 31.57,
+      "learning_rate": 6.045676767676768e-06,
+      "loss": 1.0467,
+      "step": 701000
+    },
+    {
+      "epoch": 31.6,
+      "learning_rate": 6.035575757575758e-06,
+      "loss": 1.0488,
+      "step": 701500
+    },
+    {
+      "epoch": 31.62,
+      "learning_rate": 6.025474747474748e-06,
+      "loss": 1.0497,
+      "step": 702000
+    },
+    {
+      "epoch": 31.64,
+      "learning_rate": 6.015373737373737e-06,
+      "loss": 1.0467,
+      "step": 702500
+    },
+    {
+      "epoch": 31.66,
+      "learning_rate": 6.0052727272727274e-06,
+      "loss": 1.0463,
+      "step": 703000
+    },
+    {
+      "epoch": 31.69,
+      "learning_rate": 5.99519191919192e-06,
+      "loss": 1.046,
+      "step": 703500
+    },
+    {
+      "epoch": 31.71,
+      "learning_rate": 5.985090909090909e-06,
+      "loss": 1.049,
+      "step": 704000
+    },
+    {
+      "epoch": 31.73,
+      "learning_rate": 5.974989898989899e-06,
+      "loss": 1.0443,
+      "step": 704500
+    },
+    {
+      "epoch": 31.75,
+      "learning_rate": 5.964888888888889e-06,
+      "loss": 1.0479,
+      "step": 705000
+    },
+    {
+      "epoch": 31.78,
+      "learning_rate": 5.954808080808081e-06,
+      "loss": 1.0482,
+      "step": 705500
+    },
+    {
+      "epoch": 31.8,
+      "learning_rate": 5.9447070707070705e-06,
+      "loss": 1.0462,
+      "step": 706000
+    },
+    {
+      "epoch": 31.82,
+      "learning_rate": 5.934606060606061e-06,
+      "loss": 1.0468,
+      "step": 706500
+    },
+    {
+      "epoch": 31.84,
+      "learning_rate": 5.92450505050505e-06,
+      "loss": 1.0458,
+      "step": 707000
+    },
+    {
+      "epoch": 31.87,
+      "learning_rate": 5.9144040404040406e-06,
+      "loss": 1.0453,
+      "step": 707500
+    },
+    {
+      "epoch": 31.89,
+      "learning_rate": 5.904303030303032e-06,
+      "loss": 1.0478,
+      "step": 708000
+    },
+    {
+      "epoch": 31.91,
+      "learning_rate": 5.894202020202021e-06,
+      "loss": 1.0496,
+      "step": 708500
+    },
+    {
+      "epoch": 31.93,
+      "learning_rate": 5.8841212121212125e-06,
+      "loss": 1.0488,
+      "step": 709000
+    },
+    {
+      "epoch": 31.96,
+      "learning_rate": 5.874020202020202e-06,
+      "loss": 1.0468,
+      "step": 709500
+    },
+    {
+      "epoch": 31.98,
+      "learning_rate": 5.863919191919193e-06,
+      "loss": 1.0466,
+      "step": 710000
+    },
+    {
+      "epoch": 32.0,
+      "eval_accuracy": 0.7885087042414967,
+      "eval_loss": 0.8867751955986023,
+      "eval_runtime": 54.1141,
+      "eval_samples_per_second": 903.83,
+      "eval_steps_per_second": 14.137,
+      "step": 710464
+    },
+    {
+      "epoch": 32.0,
+      "learning_rate": 5.8538181818181825e-06,
+      "loss": 1.0497,
+      "step": 710500
+    },
+    {
+      "epoch": 32.02,
+      "learning_rate": 5.843737373737374e-06,
+      "loss": 1.0475,
+      "step": 711000
+    },
+    {
+      "epoch": 32.05,
+      "learning_rate": 5.833636363636365e-06,
+      "loss": 1.0453,
+      "step": 711500
+    },
+    {
+      "epoch": 32.07,
+      "learning_rate": 5.8235353535353545e-06,
+      "loss": 1.0454,
+      "step": 712000
+    },
+    {
+      "epoch": 32.09,
+      "learning_rate": 5.813454545454545e-06,
+      "loss": 1.0469,
+      "step": 712500
+    },
+    {
+      "epoch": 32.11,
+      "learning_rate": 5.803353535353535e-06,
+      "loss": 1.0481,
+      "step": 713000
+    },
+    {
+      "epoch": 32.14,
+      "learning_rate": 5.7932525252525265e-06,
+      "loss": 1.0455,
+      "step": 713500
+    },
+    {
+      "epoch": 32.16,
+      "learning_rate": 5.783151515151516e-06,
+      "loss": 1.0475,
+      "step": 714000
+    },
+    {
+      "epoch": 32.18,
+      "learning_rate": 5.773050505050506e-06,
+      "loss": 1.0471,
+      "step": 714500
+    },
+    {
+      "epoch": 32.2,
+      "learning_rate": 5.762949494949496e-06,
+      "loss": 1.048,
+      "step": 715000
+    },
+    {
+      "epoch": 32.23,
+      "learning_rate": 5.752848484848486e-06,
+      "loss": 1.0464,
+      "step": 715500
+    },
+    {
+      "epoch": 32.25,
+      "learning_rate": 5.742747474747475e-06,
+      "loss": 1.0456,
+      "step": 716000
+    },
+    {
+      "epoch": 32.27,
+      "learning_rate": 5.732646464646465e-06,
+      "loss": 1.047,
+      "step": 716500
+    },
+    {
+      "epoch": 32.29,
+      "learning_rate": 5.722545454545455e-06,
+      "loss": 1.0483,
+      "step": 717000
+    },
+    {
+      "epoch": 32.32,
+      "learning_rate": 5.7124444444444445e-06,
+      "loss": 1.0477,
+      "step": 717500
+    },
+    {
+      "epoch": 32.34,
+      "learning_rate": 5.702343434343435e-06,
+      "loss": 1.0462,
+      "step": 718000
+    },
+    {
+      "epoch": 32.36,
+      "learning_rate": 5.692242424242424e-06,
+      "loss": 1.0461,
+      "step": 718500
+    },
+    {
+      "epoch": 32.38,
+      "learning_rate": 5.682181818181819e-06,
+      "loss": 1.0456,
+      "step": 719000
+    },
+    {
+      "epoch": 32.41,
+      "learning_rate": 5.672080808080809e-06,
+      "loss": 1.0478,
+      "step": 719500
+    },
+    {
+      "epoch": 32.43,
+      "learning_rate": 5.661979797979798e-06,
+      "loss": 1.0452,
+      "step": 720000
+    },
+    {
+      "epoch": 32.45,
+      "learning_rate": 5.6518787878787885e-06,
+      "loss": 1.0478,
+      "step": 720500
+    },
+    {
+      "epoch": 32.47,
+      "learning_rate": 5.641797979797981e-06,
+      "loss": 1.0455,
+      "step": 721000
+    },
+    {
+      "epoch": 32.5,
+      "learning_rate": 5.63169696969697e-06,
+      "loss": 1.0465,
+      "step": 721500
+    },
+    {
+      "epoch": 32.52,
+      "learning_rate": 5.6215959595959605e-06,
+      "loss": 1.0474,
+      "step": 722000
+    },
+    {
+      "epoch": 32.54,
+      "learning_rate": 5.61149494949495e-06,
+      "loss": 1.0463,
+      "step": 722500
+    },
+    {
+      "epoch": 32.56,
+      "learning_rate": 5.60139393939394e-06,
+      "loss": 1.0483,
+      "step": 723000
+    },
+    {
+      "epoch": 32.59,
+      "learning_rate": 5.59129292929293e-06,
+      "loss": 1.0458,
+      "step": 723500
+    },
+    {
+      "epoch": 32.61,
+      "learning_rate": 5.581191919191919e-06,
+      "loss": 1.0475,
+      "step": 724000
+    },
+    {
+      "epoch": 32.63,
+      "learning_rate": 5.571111111111111e-06,
+      "loss": 1.0459,
+      "step": 724500
+    },
+    {
+      "epoch": 32.65,
+      "learning_rate": 5.561010101010102e-06,
+      "loss": 1.047,
+      "step": 725000
+    },
+    {
+      "epoch": 32.68,
+      "learning_rate": 5.550909090909091e-06,
+      "loss": 1.0468,
+      "step": 725500
+    },
+    {
+      "epoch": 32.7,
+      "learning_rate": 5.540828282828283e-06,
+      "loss": 1.0457,
+      "step": 726000
+    },
+    {
+      "epoch": 32.72,
+      "learning_rate": 5.530727272727274e-06,
+      "loss": 1.0466,
+      "step": 726500
+    },
+    {
+      "epoch": 32.74,
+      "learning_rate": 5.520626262626263e-06,
+      "loss": 1.0464,
+      "step": 727000
+    },
+    {
+      "epoch": 32.77,
+      "learning_rate": 5.5105252525252525e-06,
+      "loss": 1.0465,
+      "step": 727500
+    },
+    {
+      "epoch": 32.79,
+      "learning_rate": 5.500424242424243e-06,
+      "loss": 1.0484,
+      "step": 728000
+    },
+    {
+      "epoch": 32.81,
+      "learning_rate": 5.490323232323232e-06,
+      "loss": 1.0426,
+      "step": 728500
+    },
+    {
+      "epoch": 32.83,
+      "learning_rate": 5.4802222222222225e-06,
+      "loss": 1.044,
+      "step": 729000
+    },
+    {
+      "epoch": 32.86,
+      "learning_rate": 5.470121212121212e-06,
+      "loss": 1.0465,
+      "step": 729500
+    },
+    {
+      "epoch": 32.88,
+      "learning_rate": 5.460020202020202e-06,
+      "loss": 1.045,
+      "step": 730000
+    },
+    {
+      "epoch": 32.9,
+      "learning_rate": 5.449919191919192e-06,
+      "loss": 1.0463,
+      "step": 730500
+    },
+    {
+      "epoch": 32.92,
+      "learning_rate": 5.439838383838384e-06,
+      "loss": 1.0448,
+      "step": 731000
+    },
+    {
+      "epoch": 32.95,
+      "learning_rate": 5.429737373737373e-06,
+      "loss": 1.0461,
+      "step": 731500
+    },
+    {
+      "epoch": 32.97,
+      "learning_rate": 5.419636363636364e-06,
+      "loss": 1.0457,
+      "step": 732000
+    },
+    {
+      "epoch": 32.99,
+      "learning_rate": 5.409555555555556e-06,
+      "loss": 1.0468,
+      "step": 732500
+    },
+    {
+      "epoch": 33.0,
+      "eval_accuracy": 0.7886765982693447,
+      "eval_loss": 0.8857171535491943,
+      "eval_runtime": 54.036,
+      "eval_samples_per_second": 905.138,
+      "eval_steps_per_second": 14.157,
+      "step": 732666
+    },
+    {
+      "epoch": 33.02,
+      "learning_rate": 5.399454545454545e-06,
+      "loss": 1.0458,
+      "step": 733000
+    },
+    {
+      "epoch": 33.04,
+      "learning_rate": 5.389353535353536e-06,
+      "loss": 1.0439,
+      "step": 733500
+    },
+    {
+      "epoch": 33.06,
+      "learning_rate": 5.379252525252525e-06,
+      "loss": 1.0426,
+      "step": 734000
+    },
+    {
+      "epoch": 33.08,
+      "learning_rate": 5.369151515151516e-06,
+      "loss": 1.0427,
+      "step": 734500
+    },
+    {
+      "epoch": 33.11,
+      "learning_rate": 5.359050505050506e-06,
+      "loss": 1.0468,
+      "step": 735000
+    },
+    {
+      "epoch": 33.13,
+      "learning_rate": 5.348949494949496e-06,
+      "loss": 1.0473,
+      "step": 735500
+    },
+    {
+      "epoch": 33.15,
+      "learning_rate": 5.338848484848485e-06,
+      "loss": 1.0458,
+      "step": 736000
+    },
+    {
+      "epoch": 33.17,
+      "learning_rate": 5.328767676767678e-06,
+      "loss": 1.044,
+      "step": 736500
+    },
+    {
+      "epoch": 33.2,
+      "learning_rate": 5.318666666666668e-06,
+      "loss": 1.0466,
+      "step": 737000
+    },
+    {
+      "epoch": 33.22,
+      "learning_rate": 5.308565656565657e-06,
+      "loss": 1.0475,
+      "step": 737500
+    },
+    {
+      "epoch": 33.24,
+      "learning_rate": 5.298464646464648e-06,
+      "loss": 1.0444,
+      "step": 738000
+    },
+    {
+      "epoch": 33.26,
+      "learning_rate": 5.288363636363637e-06,
+      "loss": 1.0432,
+      "step": 738500
+    },
+    {
+      "epoch": 33.29,
+      "learning_rate": 5.2782626262626265e-06,
+      "loss": 1.0461,
+      "step": 739000
+    },
+    {
+      "epoch": 33.31,
+      "learning_rate": 5.268161616161617e-06,
+      "loss": 1.0463,
+      "step": 739500
+    },
+    {
+      "epoch": 33.33,
+      "learning_rate": 5.258080808080809e-06,
+      "loss": 1.0429,
+      "step": 740000
+    },
+    {
+      "epoch": 33.35,
+      "learning_rate": 5.2479797979797984e-06,
+      "loss": 1.046,
+      "step": 740500
+    },
+    {
+      "epoch": 33.38,
+      "learning_rate": 5.237878787878789e-06,
+      "loss": 1.0417,
+      "step": 741000
+    },
+    {
+      "epoch": 33.4,
+      "learning_rate": 5.227777777777778e-06,
+      "loss": 1.0422,
+      "step": 741500
+    },
+    {
+      "epoch": 33.42,
+      "learning_rate": 5.2176767676767685e-06,
+      "loss": 1.0435,
+      "step": 742000
+    },
+    {
+      "epoch": 33.44,
+      "learning_rate": 5.20759595959596e-06,
+      "loss": 1.0436,
+      "step": 742500
+    },
+    {
+      "epoch": 33.47,
+      "learning_rate": 5.19749494949495e-06,
+      "loss": 1.0459,
+      "step": 743000
+    },
+    {
+      "epoch": 33.49,
+      "learning_rate": 5.18739393939394e-06,
+      "loss": 1.0441,
+      "step": 743500
+    },
+    {
+      "epoch": 33.51,
+      "learning_rate": 5.177313131313132e-06,
+      "loss": 1.0435,
+      "step": 744000
+    },
+    {
+      "epoch": 33.53,
+      "learning_rate": 5.167212121212122e-06,
+      "loss": 1.0439,
+      "step": 744500
+    },
+    {
+      "epoch": 33.56,
+      "learning_rate": 5.1571111111111116e-06,
+      "loss": 1.0443,
+      "step": 745000
+    },
+    {
+      "epoch": 33.58,
+      "learning_rate": 5.147010101010102e-06,
+      "loss": 1.042,
+      "step": 745500
+    },
+    {
+      "epoch": 33.6,
+      "learning_rate": 5.136929292929293e-06,
+      "loss": 1.045,
+      "step": 746000
+    },
+    {
+      "epoch": 33.62,
+      "learning_rate": 5.1268282828282835e-06,
+      "loss": 1.0442,
+      "step": 746500
+    },
+    {
+      "epoch": 33.65,
+      "learning_rate": 5.116727272727273e-06,
+      "loss": 1.044,
+      "step": 747000
+    },
+    {
+      "epoch": 33.67,
+      "learning_rate": 5.106626262626263e-06,
+      "loss": 1.0444,
+      "step": 747500
+    },
+    {
+      "epoch": 33.69,
+      "learning_rate": 5.096525252525253e-06,
+      "loss": 1.0435,
+      "step": 748000
+    },
+    {
+      "epoch": 33.71,
+      "learning_rate": 5.086424242424243e-06,
+      "loss": 1.042,
+      "step": 748500
+    },
+    {
+      "epoch": 33.74,
+      "learning_rate": 5.0763232323232324e-06,
+      "loss": 1.0469,
+      "step": 749000
+    },
+    {
+      "epoch": 33.76,
+      "learning_rate": 5.066222222222223e-06,
+      "loss": 1.0472,
+      "step": 749500
+    },
+    {
+      "epoch": 33.78,
+      "learning_rate": 5.056121212121212e-06,
+      "loss": 1.044,
+      "step": 750000
+    },
+    {
+      "epoch": 33.8,
+      "learning_rate": 5.0460202020202024e-06,
+      "loss": 1.0429,
+      "step": 750500
+    },
+    {
+      "epoch": 33.83,
+      "learning_rate": 5.035939393939394e-06,
+      "loss": 1.0437,
+      "step": 751000
+    },
+    {
+      "epoch": 33.85,
+      "learning_rate": 5.025838383838384e-06,
+      "loss": 1.0443,
+      "step": 751500
+    },
+    {
+      "epoch": 33.87,
+      "learning_rate": 5.0157373737373736e-06,
+      "loss": 1.0426,
+      "step": 752000
+    },
+    {
+      "epoch": 33.89,
+      "learning_rate": 5.005636363636364e-06,
+      "loss": 1.0451,
+      "step": 752500
+    },
+    {
+      "epoch": 33.92,
+      "learning_rate": 4.995535353535354e-06,
+      "loss": 1.0437,
+      "step": 753000
+    },
+    {
+      "epoch": 33.94,
+      "learning_rate": 4.985434343434344e-06,
+      "loss": 1.0422,
+      "step": 753500
+    },
+    {
+      "epoch": 33.96,
+      "learning_rate": 4.975333333333334e-06,
+      "loss": 1.0457,
+      "step": 754000
+    },
+    {
+      "epoch": 33.98,
+      "learning_rate": 4.965252525252526e-06,
+      "loss": 1.0464,
+      "step": 754500
+    },
+    {
+      "epoch": 34.0,
+      "eval_accuracy": 0.7891852914585747,
+      "eval_loss": 0.8822395205497742,
+      "eval_runtime": 54.2674,
+      "eval_samples_per_second": 901.277,
+      "eval_steps_per_second": 14.097,
+      "step": 754868
+    },
+    {
+      "epoch": 34.01,
+      "learning_rate": 4.9551515151515156e-06,
+      "loss": 1.043,
+      "step": 755000
+    },
+    {
+      "epoch": 34.03,
+      "learning_rate": 4.945050505050506e-06,
+      "loss": 1.0442,
+      "step": 755500
+    },
+    {
+      "epoch": 34.05,
+      "learning_rate": 4.934949494949495e-06,
+      "loss": 1.0435,
+      "step": 756000
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 4.924848484848485e-06,
+      "loss": 1.0446,
+      "step": 756500
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 4.914747474747475e-06,
+      "loss": 1.0458,
+      "step": 757000
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 4.904666666666667e-06,
+      "loss": 1.043,
+      "step": 757500
+    },
+    {
+      "epoch": 34.14,
+      "learning_rate": 4.894565656565657e-06,
+      "loss": 1.0439,
+      "step": 758000
+    },
+    {
+      "epoch": 34.16,
+      "learning_rate": 4.884484848484849e-06,
+      "loss": 1.041,
+      "step": 758500
+    },
+    {
+      "epoch": 34.19,
+      "learning_rate": 4.874383838383838e-06,
+      "loss": 1.0424,
+      "step": 759000
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 4.864282828282829e-06,
+      "loss": 1.042,
+      "step": 759500
+    },
+    {
+      "epoch": 34.23,
+      "learning_rate": 4.854181818181818e-06,
+      "loss": 1.0432,
+      "step": 760000
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 4.844080808080808e-06,
+      "loss": 1.0462,
+      "step": 760500
+    },
+    {
+      "epoch": 34.28,
+      "learning_rate": 4.833979797979798e-06,
+      "loss": 1.0446,
+      "step": 761000
+    },
+    {
+      "epoch": 34.3,
+      "learning_rate": 4.823878787878788e-06,
+      "loss": 1.0458,
+      "step": 761500
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 4.813777777777778e-06,
+      "loss": 1.0431,
+      "step": 762000
+    },
+    {
+      "epoch": 34.34,
+      "learning_rate": 4.803676767676768e-06,
+      "loss": 1.0416,
+      "step": 762500
+    },
+    {
+      "epoch": 34.37,
+      "learning_rate": 4.79359595959596e-06,
+      "loss": 1.0454,
+      "step": 763000
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 4.78349494949495e-06,
+      "loss": 1.0415,
+      "step": 763500
+    },
+    {
+      "epoch": 34.41,
+      "learning_rate": 4.77339393939394e-06,
+      "loss": 1.045,
+      "step": 764000
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 4.76329292929293e-06,
+      "loss": 1.0421,
+      "step": 764500
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 4.7532121212121215e-06,
+      "loss": 1.0427,
+      "step": 765000
+    },
+    {
+      "epoch": 34.48,
+      "learning_rate": 4.743111111111112e-06,
+      "loss": 1.0441,
+      "step": 765500
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 4.733010101010101e-06,
+      "loss": 1.0423,
+      "step": 766000
+    },
+    {
+      "epoch": 34.52,
+      "learning_rate": 4.7229292929292935e-06,
+      "loss": 1.0408,
+      "step": 766500
+    },
+    {
+      "epoch": 34.55,
+      "learning_rate": 4.712828282828284e-06,
+      "loss": 1.0447,
+      "step": 767000
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 4.702727272727273e-06,
+      "loss": 1.0433,
+      "step": 767500
+    },
+    {
+      "epoch": 34.59,
+      "learning_rate": 4.6926262626262635e-06,
+      "loss": 1.0444,
+      "step": 768000
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 4.682525252525253e-06,
+      "loss": 1.0422,
+      "step": 768500
+    },
+    {
+      "epoch": 34.64,
+      "learning_rate": 4.672424242424242e-06,
+      "loss": 1.0454,
+      "step": 769000
+    },
+    {
+      "epoch": 34.66,
+      "learning_rate": 4.662323232323233e-06,
+      "loss": 1.0436,
+      "step": 769500
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 4.652222222222222e-06,
+      "loss": 1.0437,
+      "step": 770000
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 4.642121212121212e-06,
+      "loss": 1.0443,
+      "step": 770500
+    },
+    {
+      "epoch": 34.73,
+      "learning_rate": 4.632040404040405e-06,
+      "loss": 1.0431,
+      "step": 771000
+    },
+    {
+      "epoch": 34.75,
+      "learning_rate": 4.621939393939394e-06,
+      "loss": 1.0407,
+      "step": 771500
+    },
+    {
+      "epoch": 34.77,
+      "learning_rate": 4.611838383838384e-06,
+      "loss": 1.044,
+      "step": 772000
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 4.601737373737374e-06,
+      "loss": 1.0439,
+      "step": 772500
+    },
+    {
+      "epoch": 34.82,
+      "learning_rate": 4.591636363636364e-06,
+      "loss": 1.0441,
+      "step": 773000
+    },
+    {
+      "epoch": 34.84,
+      "learning_rate": 4.581535353535354e-06,
+      "loss": 1.0446,
+      "step": 773500
+    },
+    {
+      "epoch": 34.86,
+      "learning_rate": 4.571454545454546e-06,
+      "loss": 1.0396,
+      "step": 774000
+    },
+    {
+      "epoch": 34.88,
+      "learning_rate": 4.561353535353536e-06,
+      "loss": 1.0434,
+      "step": 774500
+    },
+    {
+      "epoch": 34.91,
+      "learning_rate": 4.5512525252525255e-06,
+      "loss": 1.0427,
+      "step": 775000
+    },
+    {
+      "epoch": 34.93,
+      "learning_rate": 4.541151515151516e-06,
+      "loss": 1.0398,
+      "step": 775500
+    },
+    {
+      "epoch": 34.95,
+      "learning_rate": 4.531050505050505e-06,
+      "loss": 1.0439,
+      "step": 776000
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 4.5209494949494955e-06,
+      "loss": 1.0392,
+      "step": 776500
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 4.510848484848485e-06,
+      "loss": 1.0434,
+      "step": 777000
+    },
+    {
+      "epoch": 35.0,
+      "eval_accuracy": 0.7895888672380565,
+      "eval_loss": 0.8810570240020752,
+      "eval_runtime": 54.1595,
+      "eval_samples_per_second": 903.073,
+      "eval_steps_per_second": 14.125,
+      "step": 777070
+    },
+    {
+      "epoch": 35.02,
+      "learning_rate": 4.500747474747475e-06,
+      "loss": 1.0416,
+      "step": 777500
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 4.490646464646465e-06,
+      "loss": 1.04,
+      "step": 778000
+    },
+    {
+      "epoch": 35.06,
+      "learning_rate": 4.480565656565657e-06,
+      "loss": 1.0422,
+      "step": 778500
+    },
+    {
+      "epoch": 35.09,
+      "learning_rate": 4.470464646464646e-06,
+      "loss": 1.0415,
+      "step": 779000
+    },
+    {
+      "epoch": 35.11,
+      "learning_rate": 4.460363636363637e-06,
+      "loss": 1.0426,
+      "step": 779500
+    },
+    {
+      "epoch": 35.13,
+      "learning_rate": 4.450282828282829e-06,
+      "loss": 1.0411,
+      "step": 780000
+    },
+    {
+      "epoch": 35.15,
+      "learning_rate": 4.440181818181818e-06,
+      "loss": 1.0421,
+      "step": 780500
+    },
+    {
+      "epoch": 35.18,
+      "learning_rate": 4.430080808080809e-06,
+      "loss": 1.0435,
+      "step": 781000
+    },
+    {
+      "epoch": 35.2,
+      "learning_rate": 4.419979797979798e-06,
+      "loss": 1.0409,
+      "step": 781500
+    },
+    {
+      "epoch": 35.22,
+      "learning_rate": 4.409878787878788e-06,
+      "loss": 1.0427,
+      "step": 782000
+    },
+    {
+      "epoch": 35.24,
+      "learning_rate": 4.399777777777779e-06,
+      "loss": 1.0413,
+      "step": 782500
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 4.38969696969697e-06,
+      "loss": 1.0441,
+      "step": 783000
+    },
+    {
+      "epoch": 35.29,
+      "learning_rate": 4.3795959595959595e-06,
+      "loss": 1.0421,
+      "step": 783500
+    },
+    {
+      "epoch": 35.31,
+      "learning_rate": 4.36949494949495e-06,
+      "loss": 1.0437,
+      "step": 784000
+    },
+    {
+      "epoch": 35.33,
+      "learning_rate": 4.35939393939394e-06,
+      "loss": 1.0418,
+      "step": 784500
+    },
+    {
+      "epoch": 35.36,
+      "learning_rate": 4.3492929292929295e-06,
+      "loss": 1.0419,
+      "step": 785000
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 4.33919191919192e-06,
+      "loss": 1.0426,
+      "step": 785500
+    },
+    {
+      "epoch": 35.4,
+      "learning_rate": 4.329090909090909e-06,
+      "loss": 1.0399,
+      "step": 786000
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 4.3190101010101015e-06,
+      "loss": 1.0446,
+      "step": 786500
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 4.308909090909092e-06,
+      "loss": 1.0429,
+      "step": 787000
+    },
+    {
+      "epoch": 35.47,
+      "learning_rate": 4.298808080808081e-06,
+      "loss": 1.0436,
+      "step": 787500
+    },
+    {
+      "epoch": 35.49,
+      "learning_rate": 4.288707070707071e-06,
+      "loss": 1.0424,
+      "step": 788000
+    },
+    {
+      "epoch": 35.51,
+      "learning_rate": 4.278626262626263e-06,
+      "loss": 1.0434,
+      "step": 788500
+    },
+    {
+      "epoch": 35.54,
+      "learning_rate": 4.268525252525253e-06,
+      "loss": 1.043,
+      "step": 789000
+    },
+    {
+      "epoch": 35.56,
+      "learning_rate": 4.258424242424243e-06,
+      "loss": 1.0405,
+      "step": 789500
+    },
+    {
+      "epoch": 35.58,
+      "learning_rate": 4.248323232323233e-06,
+      "loss": 1.0431,
+      "step": 790000
+    },
+    {
+      "epoch": 35.6,
+      "learning_rate": 4.238222222222222e-06,
+      "loss": 1.0419,
+      "step": 790500
+    },
+    {
+      "epoch": 35.63,
+      "learning_rate": 4.228141414141415e-06,
+      "loss": 1.0421,
+      "step": 791000
+    },
+    {
+      "epoch": 35.65,
+      "learning_rate": 4.218040404040404e-06,
+      "loss": 1.043,
+      "step": 791500
+    },
+    {
+      "epoch": 35.67,
+      "learning_rate": 4.207939393939394e-06,
+      "loss": 1.0401,
+      "step": 792000
+    },
+    {
+      "epoch": 35.69,
+      "learning_rate": 4.197838383838384e-06,
+      "loss": 1.0432,
+      "step": 792500
+    },
+    {
+      "epoch": 35.72,
+      "learning_rate": 4.187757575757576e-06,
+      "loss": 1.0405,
+      "step": 793000
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 4.177676767676768e-06,
+      "loss": 1.0423,
+      "step": 793500
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 4.167575757575758e-06,
+      "loss": 1.0417,
+      "step": 794000
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 4.157474747474748e-06,
+      "loss": 1.0429,
+      "step": 794500
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 4.147373737373737e-06,
+      "loss": 1.039,
+      "step": 795000
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 4.137272727272728e-06,
+      "loss": 1.0399,
+      "step": 795500
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 4.127171717171717e-06,
+      "loss": 1.0415,
+      "step": 796000
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 4.1170707070707074e-06,
+      "loss": 1.0432,
+      "step": 796500
+    },
+    {
+      "epoch": 35.9,
+      "learning_rate": 4.106969696969697e-06,
+      "loss": 1.0404,
+      "step": 797000
+    },
+    {
+      "epoch": 35.92,
+      "learning_rate": 4.096868686868687e-06,
+      "loss": 1.0386,
+      "step": 797500
+    },
+    {
+      "epoch": 35.94,
+      "learning_rate": 4.0867676767676774e-06,
+      "loss": 1.0411,
+      "step": 798000
+    },
+    {
+      "epoch": 35.97,
+      "learning_rate": 4.076686868686869e-06,
+      "loss": 1.0423,
+      "step": 798500
+    },
+    {
+      "epoch": 35.99,
+      "learning_rate": 4.066585858585859e-06,
+      "loss": 1.0393,
+      "step": 799000
+    },
+    {
+      "epoch": 36.0,
+      "eval_accuracy": 0.7895544086530542,
+      "eval_loss": 0.8812911510467529,
+      "eval_runtime": 53.9501,
+      "eval_samples_per_second": 906.578,
+      "eval_steps_per_second": 14.18,
+      "step": 799272
+    },
+    {
+      "epoch": 36.01,
+      "learning_rate": 4.056484848484849e-06,
+      "loss": 1.0387,
+      "step": 799500
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 4.046383838383839e-06,
+      "loss": 1.0389,
+      "step": 800000
+    },
+    {
+      "epoch": 36.06,
+      "learning_rate": 4.036282828282828e-06,
+      "loss": 1.0424,
+      "step": 800500
+    },
+    {
+      "epoch": 36.08,
+      "learning_rate": 4.026181818181819e-06,
+      "loss": 1.0417,
+      "step": 801000
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 4.016080808080808e-06,
+      "loss": 1.0397,
+      "step": 801500
+    },
+    {
+      "epoch": 36.12,
+      "learning_rate": 4.006e-06,
+      "loss": 1.0428,
+      "step": 802000
+    },
+    {
+      "epoch": 36.15,
+      "learning_rate": 3.9958989898989906e-06,
+      "loss": 1.0388,
+      "step": 802500
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 3.98579797979798e-06,
+      "loss": 1.0409,
+      "step": 803000
+    },
+    {
+      "epoch": 36.19,
+      "learning_rate": 3.97569696969697e-06,
+      "loss": 1.04,
+      "step": 803500
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 3.96559595959596e-06,
+      "loss": 1.0376,
+      "step": 804000
+    },
+    {
+      "epoch": 36.24,
+      "learning_rate": 3.95549494949495e-06,
+      "loss": 1.041,
+      "step": 804500
+    },
+    {
+      "epoch": 36.26,
+      "learning_rate": 3.9453939393939394e-06,
+      "loss": 1.0419,
+      "step": 805000
+    },
+    {
+      "epoch": 36.28,
+      "learning_rate": 3.93529292929293e-06,
+      "loss": 1.0422,
+      "step": 805500
+    },
+    {
+      "epoch": 36.3,
+      "learning_rate": 3.925232323232324e-06,
+      "loss": 1.0402,
+      "step": 806000
+    },
+    {
+      "epoch": 36.33,
+      "learning_rate": 3.915131313131313e-06,
+      "loss": 1.0391,
+      "step": 806500
+    },
+    {
+      "epoch": 36.35,
+      "learning_rate": 3.905030303030304e-06,
+      "loss": 1.0411,
+      "step": 807000
+    },
+    {
+      "epoch": 36.37,
+      "learning_rate": 3.894929292929293e-06,
+      "loss": 1.0402,
+      "step": 807500
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 3.8848282828282826e-06,
+      "loss": 1.0381,
+      "step": 808000
+    },
+    {
+      "epoch": 36.42,
+      "learning_rate": 3.874727272727273e-06,
+      "loss": 1.0412,
+      "step": 808500
+    },
+    {
+      "epoch": 36.44,
+      "learning_rate": 3.864646464646465e-06,
+      "loss": 1.0414,
+      "step": 809000
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 3.8545454545454545e-06,
+      "loss": 1.0413,
+      "step": 809500
+    },
+    {
+      "epoch": 36.48,
+      "learning_rate": 3.844444444444445e-06,
+      "loss": 1.0423,
+      "step": 810000
+    },
+    {
+      "epoch": 36.51,
+      "learning_rate": 3.834343434343435e-06,
+      "loss": 1.039,
+      "step": 810500
+    },
+    {
+      "epoch": 36.53,
+      "learning_rate": 3.8242626262626265e-06,
+      "loss": 1.041,
+      "step": 811000
+    },
+    {
+      "epoch": 36.55,
+      "learning_rate": 3.8141616161616164e-06,
+      "loss": 1.0408,
+      "step": 811500
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 3.8040606060606062e-06,
+      "loss": 1.0395,
+      "step": 812000
+    },
+    {
+      "epoch": 36.6,
+      "learning_rate": 3.7939595959595965e-06,
+      "loss": 1.0401,
+      "step": 812500
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 3.7838585858585864e-06,
+      "loss": 1.0409,
+      "step": 813000
+    },
+    {
+      "epoch": 36.64,
+      "learning_rate": 3.7737777777777778e-06,
+      "loss": 1.0415,
+      "step": 813500
+    },
+    {
+      "epoch": 36.66,
+      "learning_rate": 3.763676767676768e-06,
+      "loss": 1.0422,
+      "step": 814000
+    },
+    {
+      "epoch": 36.69,
+      "learning_rate": 3.753575757575758e-06,
+      "loss": 1.042,
+      "step": 814500
+    },
+    {
+      "epoch": 36.71,
+      "learning_rate": 3.743474747474748e-06,
+      "loss": 1.0435,
+      "step": 815000
+    },
+    {
+      "epoch": 36.73,
+      "learning_rate": 3.7333737373737377e-06,
+      "loss": 1.0431,
+      "step": 815500
+    },
+    {
+      "epoch": 36.75,
+      "learning_rate": 3.72329292929293e-06,
+      "loss": 1.0379,
+      "step": 816000
+    },
+    {
+      "epoch": 36.78,
+      "learning_rate": 3.7131919191919198e-06,
+      "loss": 1.0392,
+      "step": 816500
+    },
+    {
+      "epoch": 36.8,
+      "learning_rate": 3.7030909090909096e-06,
+      "loss": 1.0412,
+      "step": 817000
+    },
+    {
+      "epoch": 36.82,
+      "learning_rate": 3.6929898989898995e-06,
+      "loss": 1.0385,
+      "step": 817500
+    },
+    {
+      "epoch": 36.84,
+      "learning_rate": 3.6828888888888894e-06,
+      "loss": 1.0395,
+      "step": 818000
+    },
+    {
+      "epoch": 36.87,
+      "learning_rate": 3.6727878787878792e-06,
+      "loss": 1.038,
+      "step": 818500
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 3.6626868686868687e-06,
+      "loss": 1.0408,
+      "step": 819000
+    },
+    {
+      "epoch": 36.91,
+      "learning_rate": 3.652606060606061e-06,
+      "loss": 1.0392,
+      "step": 819500
+    },
+    {
+      "epoch": 36.93,
+      "learning_rate": 3.6425050505050508e-06,
+      "loss": 1.0409,
+      "step": 820000
+    },
+    {
+      "epoch": 36.96,
+      "learning_rate": 3.6324040404040406e-06,
+      "loss": 1.0403,
+      "step": 820500
+    },
+    {
+      "epoch": 36.98,
+      "learning_rate": 3.6223030303030305e-06,
+      "loss": 1.0412,
+      "step": 821000
+    },
+    {
+      "epoch": 37.0,
+      "eval_accuracy": 0.7899152841708073,
+      "eval_loss": 0.8797088861465454,
+      "eval_runtime": 59.8662,
+      "eval_samples_per_second": 816.989,
+      "eval_steps_per_second": 12.779,
+      "step": 821474
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 3.6122020202020204e-06,
+      "loss": 1.0384,
+      "step": 821500
+    },
+    {
+      "epoch": 37.02,
+      "learning_rate": 3.6021010101010106e-06,
+      "loss": 1.0397,
+      "step": 822000
+    },
+    {
+      "epoch": 37.05,
+      "learning_rate": 3.5920000000000005e-06,
+      "loss": 1.0406,
+      "step": 822500
+    },
+    {
+      "epoch": 37.07,
+      "learning_rate": 3.5818989898989904e-06,
+      "loss": 1.0386,
+      "step": 823000
+    },
+    {
+      "epoch": 37.09,
+      "learning_rate": 3.5718181818181818e-06,
+      "loss": 1.044,
+      "step": 823500
+    },
+    {
+      "epoch": 37.11,
+      "learning_rate": 3.561717171717172e-06,
+      "loss": 1.0373,
+      "step": 824000
+    },
+    {
+      "epoch": 37.14,
+      "learning_rate": 3.551616161616162e-06,
+      "loss": 1.0407,
+      "step": 824500
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 3.5415353535353537e-06,
+      "loss": 1.0401,
+      "step": 825000
+    },
+    {
+      "epoch": 37.18,
+      "learning_rate": 3.531434343434344e-06,
+      "loss": 1.0388,
+      "step": 825500
+    },
+    {
+      "epoch": 37.2,
+      "learning_rate": 3.521333333333334e-06,
+      "loss": 1.0392,
+      "step": 826000
+    },
+    {
+      "epoch": 37.23,
+      "learning_rate": 3.5112323232323238e-06,
+      "loss": 1.039,
+      "step": 826500
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 3.501151515151515e-06,
+      "loss": 1.0396,
+      "step": 827000
+    },
+    {
+      "epoch": 37.27,
+      "learning_rate": 3.4910505050505054e-06,
+      "loss": 1.039,
+      "step": 827500
+    },
+    {
+      "epoch": 37.29,
+      "learning_rate": 3.4809494949494953e-06,
+      "loss": 1.0392,
+      "step": 828000
+    },
+    {
+      "epoch": 37.32,
+      "learning_rate": 3.470848484848485e-06,
+      "loss": 1.0406,
+      "step": 828500
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 3.460747474747475e-06,
+      "loss": 1.0405,
+      "step": 829000
+    },
+    {
+      "epoch": 37.36,
+      "learning_rate": 3.450646464646465e-06,
+      "loss": 1.0396,
+      "step": 829500
+    },
+    {
+      "epoch": 37.38,
+      "learning_rate": 3.440565656565657e-06,
+      "loss": 1.0414,
+      "step": 830000
+    },
+    {
+      "epoch": 37.41,
+      "learning_rate": 3.430464646464647e-06,
+      "loss": 1.0391,
+      "step": 830500
+    },
+    {
+      "epoch": 37.43,
+      "learning_rate": 3.420363636363637e-06,
+      "loss": 1.041,
+      "step": 831000
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 3.4102626262626263e-06,
+      "loss": 1.0407,
+      "step": 831500
+    },
+    {
+      "epoch": 37.47,
+      "learning_rate": 3.400161616161616e-06,
+      "loss": 1.0384,
+      "step": 832000
+    },
+    {
+      "epoch": 37.5,
+      "learning_rate": 3.3900808080808084e-06,
+      "loss": 1.0381,
+      "step": 832500
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 3.3799797979797983e-06,
+      "loss": 1.0379,
+      "step": 833000
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 3.369878787878788e-06,
+      "loss": 1.0389,
+      "step": 833500
+    },
+    {
+      "epoch": 37.56,
+      "learning_rate": 3.359777777777778e-06,
+      "loss": 1.0383,
+      "step": 834000
+    },
+    {
+      "epoch": 37.59,
+      "learning_rate": 3.349676767676768e-06,
+      "loss": 1.0384,
+      "step": 834500
+    },
+    {
+      "epoch": 37.61,
+      "learning_rate": 3.3395959595959597e-06,
+      "loss": 1.0385,
+      "step": 835000
+    },
+    {
+      "epoch": 37.63,
+      "learning_rate": 3.3294949494949496e-06,
+      "loss": 1.0368,
+      "step": 835500
+    },
+    {
+      "epoch": 37.65,
+      "learning_rate": 3.3193939393939394e-06,
+      "loss": 1.039,
+      "step": 836000
+    },
+    {
+      "epoch": 37.68,
+      "learning_rate": 3.3092929292929293e-06,
+      "loss": 1.0372,
+      "step": 836500
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 3.2992121212121215e-06,
+      "loss": 1.0381,
+      "step": 837000
+    },
+    {
+      "epoch": 37.72,
+      "learning_rate": 3.2891111111111114e-06,
+      "loss": 1.0386,
+      "step": 837500
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 3.2790101010101013e-06,
+      "loss": 1.0379,
+      "step": 838000
+    },
+    {
+      "epoch": 37.77,
+      "learning_rate": 3.268909090909091e-06,
+      "loss": 1.0382,
+      "step": 838500
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 3.2588080808080814e-06,
+      "loss": 1.0421,
+      "step": 839000
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 3.2487070707070713e-06,
+      "loss": 1.0384,
+      "step": 839500
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 3.2386262626262627e-06,
+      "loss": 1.0393,
+      "step": 840000
+    },
+    {
+      "epoch": 37.86,
+      "learning_rate": 3.228525252525253e-06,
+      "loss": 1.041,
+      "step": 840500
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 3.218444444444445e-06,
+      "loss": 1.0388,
+      "step": 841000
+    },
+    {
+      "epoch": 37.9,
+      "learning_rate": 3.2083434343434342e-06,
+      "loss": 1.0408,
+      "step": 841500
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 3.198242424242424e-06,
+      "loss": 1.0396,
+      "step": 842000
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 3.188141414141415e-06,
+      "loss": 1.0392,
+      "step": 842500
+    },
+    {
+      "epoch": 37.97,
+      "learning_rate": 3.1780404040404047e-06,
+      "loss": 1.0389,
+      "step": 843000
+    },
+    {
+      "epoch": 37.99,
+      "learning_rate": 3.167939393939394e-06,
+      "loss": 1.04,
+      "step": 843500
+    },
+    {
+      "epoch": 38.0,
+      "eval_accuracy": 0.7894997585529906,
+      "eval_loss": 0.8806864619255066,
+      "eval_runtime": 59.8351,
+      "eval_samples_per_second": 817.414,
+      "eval_steps_per_second": 12.785,
+      "step": 843676
+    },
+    {
+      "epoch": 38.01,
+      "learning_rate": 3.157838383838384e-06,
+      "loss": 1.0395,
+      "step": 844000
+    },
+    {
+      "epoch": 38.04,
+      "learning_rate": 3.147737373737374e-06,
+      "loss": 1.0378,
+      "step": 844500
+    },
+    {
+      "epoch": 38.06,
+      "learning_rate": 3.137656565656566e-06,
+      "loss": 1.0403,
+      "step": 845000
+    },
+    {
+      "epoch": 38.08,
+      "learning_rate": 3.127555555555556e-06,
+      "loss": 1.0394,
+      "step": 845500
+    },
+    {
+      "epoch": 38.1,
+      "learning_rate": 3.117454545454546e-06,
+      "loss": 1.0378,
+      "step": 846000
+    },
+    {
+      "epoch": 38.13,
+      "learning_rate": 3.1073535353535357e-06,
+      "loss": 1.0391,
+      "step": 846500
+    },
+    {
+      "epoch": 38.15,
+      "learning_rate": 3.0972525252525255e-06,
+      "loss": 1.0369,
+      "step": 847000
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 3.0871717171717174e-06,
+      "loss": 1.0395,
+      "step": 847500
+    },
+    {
+      "epoch": 38.19,
+      "learning_rate": 3.0770707070707072e-06,
+      "loss": 1.0411,
+      "step": 848000
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 3.066969696969697e-06,
+      "loss": 1.0381,
+      "step": 848500
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 3.056868686868687e-06,
+      "loss": 1.0389,
+      "step": 849000
+    },
+    {
+      "epoch": 38.26,
+      "learning_rate": 3.046767676767677e-06,
+      "loss": 1.0368,
+      "step": 849500
+    },
+    {
+      "epoch": 38.28,
+      "learning_rate": 3.0366666666666667e-06,
+      "loss": 1.0387,
+      "step": 850000
+    },
+    {
+      "epoch": 38.31,
+      "learning_rate": 3.026565656565657e-06,
+      "loss": 1.0375,
+      "step": 850500
+    },
+    {
+      "epoch": 38.33,
+      "learning_rate": 3.0164848484848484e-06,
+      "loss": 1.0361,
+      "step": 851000
+    },
+    {
+      "epoch": 38.35,
+      "learning_rate": 3.0063838383838382e-06,
+      "loss": 1.0404,
+      "step": 851500
+    },
+    {
+      "epoch": 38.37,
+      "learning_rate": 2.9963030303030305e-06,
+      "loss": 1.0386,
+      "step": 852000
+    },
+    {
+      "epoch": 38.4,
+      "learning_rate": 2.9862020202020203e-06,
+      "loss": 1.0393,
+      "step": 852500
+    },
+    {
+      "epoch": 38.42,
+      "learning_rate": 2.97610101010101e-06,
+      "loss": 1.0368,
+      "step": 853000
+    },
+    {
+      "epoch": 38.44,
+      "learning_rate": 2.966e-06,
+      "loss": 1.0395,
+      "step": 853500
+    },
+    {
+      "epoch": 38.47,
+      "learning_rate": 2.9558989898989903e-06,
+      "loss": 1.0372,
+      "step": 854000
+    },
+    {
+      "epoch": 38.49,
+      "learning_rate": 2.9457979797979802e-06,
+      "loss": 1.0374,
+      "step": 854500
+    },
+    {
+      "epoch": 38.51,
+      "learning_rate": 2.93569696969697e-06,
+      "loss": 1.037,
+      "step": 855000
+    },
+    {
+      "epoch": 38.53,
+      "learning_rate": 2.92559595959596e-06,
+      "loss": 1.0388,
+      "step": 855500
+    },
+    {
+      "epoch": 38.56,
+      "learning_rate": 2.91549494949495e-06,
+      "loss": 1.037,
+      "step": 856000
+    },
+    {
+      "epoch": 38.58,
+      "learning_rate": 2.9054141414141416e-06,
+      "loss": 1.0375,
+      "step": 856500
+    },
+    {
+      "epoch": 38.6,
+      "learning_rate": 2.8953131313131315e-06,
+      "loss": 1.0385,
+      "step": 857000
+    },
+    {
+      "epoch": 38.62,
+      "learning_rate": 2.8852121212121214e-06,
+      "loss": 1.0378,
+      "step": 857500
+    },
+    {
+      "epoch": 38.65,
+      "learning_rate": 2.8751111111111112e-06,
+      "loss": 1.037,
+      "step": 858000
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 2.865010101010101e-06,
+      "loss": 1.0376,
+      "step": 858500
+    },
+    {
+      "epoch": 38.69,
+      "learning_rate": 2.854909090909091e-06,
+      "loss": 1.0376,
+      "step": 859000
+    },
+    {
+      "epoch": 38.71,
+      "learning_rate": 2.844808080808081e-06,
+      "loss": 1.0391,
+      "step": 859500
+    },
+    {
+      "epoch": 38.74,
+      "learning_rate": 2.834707070707071e-06,
+      "loss": 1.0387,
+      "step": 860000
+    },
+    {
+      "epoch": 38.76,
+      "learning_rate": 2.824626262626263e-06,
+      "loss": 1.0398,
+      "step": 860500
+    },
+    {
+      "epoch": 38.78,
+      "learning_rate": 2.8145252525252524e-06,
+      "loss": 1.0384,
+      "step": 861000
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 2.8044242424242422e-06,
+      "loss": 1.0346,
+      "step": 861500
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 2.7943434343434345e-06,
+      "loss": 1.0393,
+      "step": 862000
+    },
+    {
+      "epoch": 38.85,
+      "learning_rate": 2.7842424242424243e-06,
+      "loss": 1.0374,
+      "step": 862500
+    },
+    {
+      "epoch": 38.87,
+      "learning_rate": 2.774141414141414e-06,
+      "loss": 1.0409,
+      "step": 863000
+    },
+    {
+      "epoch": 38.89,
+      "learning_rate": 2.7640404040404045e-06,
+      "loss": 1.0404,
+      "step": 863500
+    },
+    {
+      "epoch": 38.92,
+      "learning_rate": 2.7539393939393943e-06,
+      "loss": 1.0395,
+      "step": 864000
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 2.7438585858585857e-06,
+      "loss": 1.0351,
+      "step": 864500
+    },
+    {
+      "epoch": 38.96,
+      "learning_rate": 2.7337575757575756e-06,
+      "loss": 1.038,
+      "step": 865000
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 2.7236565656565663e-06,
+      "loss": 1.0379,
+      "step": 865500
+    },
+    {
+      "epoch": 39.0,
+      "eval_accuracy": 0.7901029189779466,
+      "eval_loss": 0.8793747425079346,
+      "eval_runtime": 59.8765,
+      "eval_samples_per_second": 816.847,
+      "eval_steps_per_second": 12.776,
+      "step": 865878
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 2.7135555555555558e-06,
+      "loss": 1.0363,
+      "step": 866000
+    },
+    {
+      "epoch": 39.03,
+      "learning_rate": 2.7034545454545456e-06,
+      "loss": 1.0356,
+      "step": 866500
+    },
+    {
+      "epoch": 39.05,
+      "learning_rate": 2.6933535353535355e-06,
+      "loss": 1.039,
+      "step": 867000
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 2.6832525252525253e-06,
+      "loss": 1.0397,
+      "step": 867500
+    },
+    {
+      "epoch": 39.1,
+      "learning_rate": 2.6731717171717176e-06,
+      "loss": 1.0371,
+      "step": 868000
+    },
+    {
+      "epoch": 39.12,
+      "learning_rate": 2.6630707070707075e-06,
+      "loss": 1.0376,
+      "step": 868500
+    },
+    {
+      "epoch": 39.14,
+      "learning_rate": 2.6529696969696973e-06,
+      "loss": 1.0383,
+      "step": 869000
+    },
+    {
+      "epoch": 39.16,
+      "learning_rate": 2.642868686868687e-06,
+      "loss": 1.035,
+      "step": 869500
+    },
+    {
+      "epoch": 39.19,
+      "learning_rate": 2.632767676767677e-06,
+      "loss": 1.0363,
+      "step": 870000
+    },
+    {
+      "epoch": 39.21,
+      "learning_rate": 2.622686868686869e-06,
+      "loss": 1.0376,
+      "step": 870500
+    },
+    {
+      "epoch": 39.23,
+      "learning_rate": 2.612606060606061e-06,
+      "loss": 1.0376,
+      "step": 871000
+    },
+    {
+      "epoch": 39.25,
+      "learning_rate": 2.602505050505051e-06,
+      "loss": 1.0378,
+      "step": 871500
+    },
+    {
+      "epoch": 39.28,
+      "learning_rate": 2.592404040404041e-06,
+      "loss": 1.0372,
+      "step": 872000
+    },
+    {
+      "epoch": 39.3,
+      "learning_rate": 2.5823030303030307e-06,
+      "loss": 1.0389,
+      "step": 872500
+    },
+    {
+      "epoch": 39.32,
+      "learning_rate": 2.5722020202020206e-06,
+      "loss": 1.0373,
+      "step": 873000
+    },
+    {
+      "epoch": 39.34,
+      "learning_rate": 2.56210101010101e-06,
+      "loss": 1.0355,
+      "step": 873500
+    },
+    {
+      "epoch": 39.37,
+      "learning_rate": 2.5520202020202023e-06,
+      "loss": 1.0395,
+      "step": 874000
+    },
+    {
+      "epoch": 39.39,
+      "learning_rate": 2.541919191919192e-06,
+      "loss": 1.0379,
+      "step": 874500
+    },
+    {
+      "epoch": 39.41,
+      "learning_rate": 2.531818181818182e-06,
+      "loss": 1.0367,
+      "step": 875000
+    },
+    {
+      "epoch": 39.43,
+      "learning_rate": 2.521717171717172e-06,
+      "loss": 1.0395,
+      "step": 875500
+    },
+    {
+      "epoch": 39.46,
+      "learning_rate": 2.5116161616161617e-06,
+      "loss": 1.0369,
+      "step": 876000
+    },
+    {
+      "epoch": 39.48,
+      "learning_rate": 2.5015151515151516e-06,
+      "loss": 1.0363,
+      "step": 876500
+    },
+    {
+      "epoch": 39.5,
+      "learning_rate": 2.4914141414141414e-06,
+      "loss": 1.036,
+      "step": 877000
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 2.4813131313131313e-06,
+      "loss": 1.0372,
+      "step": 877500
+    },
+    {
+      "epoch": 39.55,
+      "learning_rate": 2.4712121212121216e-06,
+      "loss": 1.037,
+      "step": 878000
+    },
+    {
+      "epoch": 39.57,
+      "learning_rate": 2.4611515151515154e-06,
+      "loss": 1.0381,
+      "step": 878500
+    },
+    {
+      "epoch": 39.59,
+      "learning_rate": 2.4510505050505052e-06,
+      "loss": 1.0388,
+      "step": 879000
+    },
+    {
+      "epoch": 39.61,
+      "learning_rate": 2.440949494949495e-06,
+      "loss": 1.0355,
+      "step": 879500
+    },
+    {
+      "epoch": 39.64,
+      "learning_rate": 2.430848484848485e-06,
+      "loss": 1.0359,
+      "step": 880000
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 2.420747474747475e-06,
+      "loss": 1.038,
+      "step": 880500
+    },
+    {
+      "epoch": 39.68,
+      "learning_rate": 2.4106464646464647e-06,
+      "loss": 1.0382,
+      "step": 881000
+    },
+    {
+      "epoch": 39.7,
+      "learning_rate": 2.400545454545455e-06,
+      "loss": 1.0361,
+      "step": 881500
+    },
+    {
+      "epoch": 39.73,
+      "learning_rate": 2.390444444444445e-06,
+      "loss": 1.0374,
+      "step": 882000
+    },
+    {
+      "epoch": 39.75,
+      "learning_rate": 2.3803434343434347e-06,
+      "loss": 1.0347,
+      "step": 882500
+    },
+    {
+      "epoch": 39.77,
+      "learning_rate": 2.370242424242424e-06,
+      "loss": 1.0358,
+      "step": 883000
+    },
+    {
+      "epoch": 39.79,
+      "learning_rate": 2.3601414141414144e-06,
+      "loss": 1.0363,
+      "step": 883500
+    },
+    {
+      "epoch": 39.82,
+      "learning_rate": 2.3500404040404043e-06,
+      "loss": 1.0379,
+      "step": 884000
+    },
+    {
+      "epoch": 39.84,
+      "learning_rate": 2.339959595959596e-06,
+      "loss": 1.0358,
+      "step": 884500
+    },
+    {
+      "epoch": 39.86,
+      "learning_rate": 2.3298787878787884e-06,
+      "loss": 1.0378,
+      "step": 885000
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 2.319777777777778e-06,
+      "loss": 1.0392,
+      "step": 885500
+    },
+    {
+      "epoch": 39.91,
+      "learning_rate": 2.3096767676767677e-06,
+      "loss": 1.0378,
+      "step": 886000
+    },
+    {
+      "epoch": 39.93,
+      "learning_rate": 2.2995959595959595e-06,
+      "loss": 1.0357,
+      "step": 886500
+    },
+    {
+      "epoch": 39.95,
+      "learning_rate": 2.2894949494949498e-06,
+      "loss": 1.0371,
+      "step": 887000
+    },
+    {
+      "epoch": 39.97,
+      "learning_rate": 2.2793939393939396e-06,
+      "loss": 1.0394,
+      "step": 887500
+    },
+    {
+      "epoch": 40.0,
+      "learning_rate": 2.2692929292929295e-06,
+      "loss": 1.0379,
+      "step": 888000
+    },
+    {
+      "epoch": 40.0,
+      "eval_accuracy": 0.7900979663838801,
+      "eval_loss": 0.8782429099082947,
+      "eval_runtime": 59.7918,
+      "eval_samples_per_second": 818.005,
+      "eval_steps_per_second": 12.794,
+      "step": 888080
+    },
+    {
+      "epoch": 40.02,
+      "learning_rate": 2.2591919191919194e-06,
+      "loss": 1.0365,
+      "step": 888500
+    },
+    {
+      "epoch": 40.04,
+      "learning_rate": 2.2490909090909092e-06,
+      "loss": 1.0361,
+      "step": 889000
+    },
+    {
+      "epoch": 40.06,
+      "learning_rate": 2.238989898989899e-06,
+      "loss": 1.0344,
+      "step": 889500
+    },
+    {
+      "epoch": 40.09,
+      "learning_rate": 2.228888888888889e-06,
+      "loss": 1.0368,
+      "step": 890000
+    },
+    {
+      "epoch": 40.11,
+      "learning_rate": 2.2188080808080808e-06,
+      "loss": 1.0381,
+      "step": 890500
+    },
+    {
+      "epoch": 40.13,
+      "learning_rate": 2.208707070707071e-06,
+      "loss": 1.038,
+      "step": 891000
+    },
+    {
+      "epoch": 40.15,
+      "learning_rate": 2.198606060606061e-06,
+      "loss": 1.0373,
+      "step": 891500
+    },
+    {
+      "epoch": 40.18,
+      "learning_rate": 2.1885252525252528e-06,
+      "loss": 1.0359,
+      "step": 892000
+    },
+    {
+      "epoch": 40.2,
+      "learning_rate": 2.1784242424242426e-06,
+      "loss": 1.0352,
+      "step": 892500
+    },
+    {
+      "epoch": 40.22,
+      "learning_rate": 2.1683232323232325e-06,
+      "loss": 1.0357,
+      "step": 893000
+    },
+    {
+      "epoch": 40.24,
+      "learning_rate": 2.1582222222222223e-06,
+      "loss": 1.0391,
+      "step": 893500
+    },
+    {
+      "epoch": 40.27,
+      "learning_rate": 2.148121212121212e-06,
+      "loss": 1.0364,
+      "step": 894000
+    },
+    {
+      "epoch": 40.29,
+      "learning_rate": 2.1380202020202025e-06,
+      "loss": 1.0355,
+      "step": 894500
+    },
+    {
+      "epoch": 40.31,
+      "learning_rate": 2.1279393939393943e-06,
+      "loss": 1.0354,
+      "step": 895000
+    },
+    {
+      "epoch": 40.33,
+      "learning_rate": 2.1178383838383838e-06,
+      "loss": 1.0374,
+      "step": 895500
+    },
+    {
+      "epoch": 40.36,
+      "learning_rate": 2.1077373737373736e-06,
+      "loss": 1.037,
+      "step": 896000
+    },
+    {
+      "epoch": 40.38,
+      "learning_rate": 2.097636363636364e-06,
+      "loss": 1.0372,
+      "step": 896500
+    },
+    {
+      "epoch": 40.4,
+      "learning_rate": 2.0875353535353538e-06,
+      "loss": 1.036,
+      "step": 897000
+    },
+    {
+      "epoch": 40.42,
+      "learning_rate": 2.0774545454545456e-06,
+      "loss": 1.037,
+      "step": 897500
+    },
+    {
+      "epoch": 40.45,
+      "learning_rate": 2.0673535353535355e-06,
+      "loss": 1.0368,
+      "step": 898000
+    },
+    {
+      "epoch": 40.47,
+      "learning_rate": 2.0572525252525253e-06,
+      "loss": 1.0385,
+      "step": 898500
+    },
+    {
+      "epoch": 40.49,
+      "learning_rate": 2.047151515151515e-06,
+      "loss": 1.0348,
+      "step": 899000
+    },
+    {
+      "epoch": 40.51,
+      "learning_rate": 2.037050505050505e-06,
+      "loss": 1.036,
+      "step": 899500
+    },
+    {
+      "epoch": 40.54,
+      "learning_rate": 2.026949494949495e-06,
+      "loss": 1.0344,
+      "step": 900000
+    },
+    {
+      "epoch": 40.56,
+      "learning_rate": 2.016848484848485e-06,
+      "loss": 1.0366,
+      "step": 900500
+    },
+    {
+      "epoch": 40.58,
+      "learning_rate": 2.006747474747475e-06,
+      "loss": 1.0349,
+      "step": 901000
+    },
+    {
+      "epoch": 40.6,
+      "learning_rate": 1.996646464646465e-06,
+      "loss": 1.0336,
+      "step": 901500
+    },
+    {
+      "epoch": 40.63,
+      "learning_rate": 1.986545454545455e-06,
+      "loss": 1.0364,
+      "step": 902000
+    },
+    {
+      "epoch": 40.65,
+      "learning_rate": 1.9764444444444447e-06,
+      "loss": 1.0363,
+      "step": 902500
+    },
+    {
+      "epoch": 40.67,
+      "learning_rate": 1.9663636363636365e-06,
+      "loss": 1.0353,
+      "step": 903000
+    },
+    {
+      "epoch": 40.69,
+      "learning_rate": 1.9562626262626263e-06,
+      "loss": 1.0354,
+      "step": 903500
+    },
+    {
+      "epoch": 40.72,
+      "learning_rate": 1.946161616161616e-06,
+      "loss": 1.0371,
+      "step": 904000
+    },
+    {
+      "epoch": 40.74,
+      "learning_rate": 1.9360606060606065e-06,
+      "loss": 1.0356,
+      "step": 904500
+    },
+    {
+      "epoch": 40.76,
+      "learning_rate": 1.925959595959596e-06,
+      "loss": 1.0356,
+      "step": 905000
+    },
+    {
+      "epoch": 40.78,
+      "learning_rate": 1.915858585858586e-06,
+      "loss": 1.0363,
+      "step": 905500
+    },
+    {
+      "epoch": 40.81,
+      "learning_rate": 1.9057575757575759e-06,
+      "loss": 1.039,
+      "step": 906000
+    },
+    {
+      "epoch": 40.83,
+      "learning_rate": 1.895656565656566e-06,
+      "loss": 1.0367,
+      "step": 906500
+    },
+    {
+      "epoch": 40.85,
+      "learning_rate": 1.8855757575757578e-06,
+      "loss": 1.0351,
+      "step": 907000
+    },
+    {
+      "epoch": 40.87,
+      "learning_rate": 1.8754747474747476e-06,
+      "loss": 1.037,
+      "step": 907500
+    },
+    {
+      "epoch": 40.9,
+      "learning_rate": 1.8653737373737375e-06,
+      "loss": 1.0353,
+      "step": 908000
+    },
+    {
+      "epoch": 40.92,
+      "learning_rate": 1.8552727272727276e-06,
+      "loss": 1.034,
+      "step": 908500
+    },
+    {
+      "epoch": 40.94,
+      "learning_rate": 1.8451919191919194e-06,
+      "loss": 1.0362,
+      "step": 909000
+    },
+    {
+      "epoch": 40.96,
+      "learning_rate": 1.8351111111111114e-06,
+      "loss": 1.0346,
+      "step": 909500
+    },
+    {
+      "epoch": 40.99,
+      "learning_rate": 1.8250101010101013e-06,
+      "loss": 1.0396,
+      "step": 910000
+    },
+    {
+      "epoch": 41.0,
+      "eval_accuracy": 0.7904704041711276,
+      "eval_loss": 0.8766286969184875,
+      "eval_runtime": 53.4936,
+      "eval_samples_per_second": 914.316,
+      "eval_steps_per_second": 14.301,
+      "step": 910282
+    },
+    {
+      "epoch": 41.01,
+      "learning_rate": 1.8149292929292931e-06,
+      "loss": 1.0349,
+      "step": 910500
+    },
+    {
+      "epoch": 41.03,
+      "learning_rate": 1.8048282828282828e-06,
+      "loss": 1.0361,
+      "step": 911000
+    },
+    {
+      "epoch": 41.05,
+      "learning_rate": 1.794727272727273e-06,
+      "loss": 1.0344,
+      "step": 911500
+    },
+    {
+      "epoch": 41.08,
+      "learning_rate": 1.7846464646464649e-06,
+      "loss": 1.0368,
+      "step": 912000
+    },
+    {
+      "epoch": 41.1,
+      "learning_rate": 1.7745454545454545e-06,
+      "loss": 1.0369,
+      "step": 912500
+    },
+    {
+      "epoch": 41.12,
+      "learning_rate": 1.7644444444444444e-06,
+      "loss": 1.0335,
+      "step": 913000
+    },
+    {
+      "epoch": 41.14,
+      "learning_rate": 1.7543434343434345e-06,
+      "loss": 1.0354,
+      "step": 913500
+    },
+    {
+      "epoch": 41.17,
+      "learning_rate": 1.7442424242424243e-06,
+      "loss": 1.035,
+      "step": 914000
+    },
+    {
+      "epoch": 41.19,
+      "learning_rate": 1.7341414141414142e-06,
+      "loss": 1.037,
+      "step": 914500
+    },
+    {
+      "epoch": 41.21,
+      "learning_rate": 1.724040404040404e-06,
+      "loss": 1.0337,
+      "step": 915000
+    },
+    {
+      "epoch": 41.24,
+      "learning_rate": 1.7139393939393941e-06,
+      "loss": 1.0362,
+      "step": 915500
+    },
+    {
+      "epoch": 41.26,
+      "learning_rate": 1.703858585858586e-06,
+      "loss": 1.0391,
+      "step": 916000
+    },
+    {
+      "epoch": 41.28,
+      "learning_rate": 1.6937575757575758e-06,
+      "loss": 1.0339,
+      "step": 916500
+    },
+    {
+      "epoch": 41.3,
+      "learning_rate": 1.6836565656565657e-06,
+      "loss": 1.0376,
+      "step": 917000
+    },
+    {
+      "epoch": 41.33,
+      "learning_rate": 1.6735555555555558e-06,
+      "loss": 1.0351,
+      "step": 917500
+    },
+    {
+      "epoch": 41.35,
+      "learning_rate": 1.6634545454545456e-06,
+      "loss": 1.0346,
+      "step": 918000
+    },
+    {
+      "epoch": 41.37,
+      "learning_rate": 1.6533535353535355e-06,
+      "loss": 1.0397,
+      "step": 918500
+    },
+    {
+      "epoch": 41.39,
+      "learning_rate": 1.6432525252525254e-06,
+      "loss": 1.0343,
+      "step": 919000
+    },
+    {
+      "epoch": 41.42,
+      "learning_rate": 1.6331515151515154e-06,
+      "loss": 1.034,
+      "step": 919500
+    },
+    {
+      "epoch": 41.44,
+      "learning_rate": 1.6230707070707073e-06,
+      "loss": 1.0353,
+      "step": 920000
+    },
+    {
+      "epoch": 41.46,
+      "learning_rate": 1.6129696969696971e-06,
+      "loss": 1.0365,
+      "step": 920500
+    },
+    {
+      "epoch": 41.48,
+      "learning_rate": 1.6028686868686872e-06,
+      "loss": 1.0344,
+      "step": 921000
+    },
+    {
+      "epoch": 41.51,
+      "learning_rate": 1.592767676767677e-06,
+      "loss": 1.0375,
+      "step": 921500
+    },
+    {
+      "epoch": 41.53,
+      "learning_rate": 1.5826666666666667e-06,
+      "loss": 1.0336,
+      "step": 922000
+    },
+    {
+      "epoch": 41.55,
+      "learning_rate": 1.5725656565656566e-06,
+      "loss": 1.0326,
+      "step": 922500
+    },
+    {
+      "epoch": 41.57,
+      "learning_rate": 1.5624646464646464e-06,
+      "loss": 1.0356,
+      "step": 923000
+    },
+    {
+      "epoch": 41.6,
+      "learning_rate": 1.5523636363636365e-06,
+      "loss": 1.0351,
+      "step": 923500
+    },
+    {
+      "epoch": 41.62,
+      "learning_rate": 1.5422626262626264e-06,
+      "loss": 1.0341,
+      "step": 924000
+    },
+    {
+      "epoch": 41.64,
+      "learning_rate": 1.5321818181818182e-06,
+      "loss": 1.0353,
+      "step": 924500
+    },
+    {
+      "epoch": 41.66,
+      "learning_rate": 1.5221010101010102e-06,
+      "loss": 1.0374,
+      "step": 925000
+    },
+    {
+      "epoch": 41.69,
+      "learning_rate": 1.512e-06,
+      "loss": 1.0348,
+      "step": 925500
+    },
+    {
+      "epoch": 41.71,
+      "learning_rate": 1.50189898989899e-06,
+      "loss": 1.034,
+      "step": 926000
+    },
+    {
+      "epoch": 41.73,
+      "learning_rate": 1.4917979797979798e-06,
+      "loss": 1.0356,
+      "step": 926500
+    },
+    {
+      "epoch": 41.75,
+      "learning_rate": 1.4817171717171719e-06,
+      "loss": 1.0362,
+      "step": 927000
+    },
+    {
+      "epoch": 41.78,
+      "learning_rate": 1.4716161616161617e-06,
+      "loss": 1.0329,
+      "step": 927500
+    },
+    {
+      "epoch": 41.8,
+      "learning_rate": 1.4615151515151516e-06,
+      "loss": 1.0343,
+      "step": 928000
+    },
+    {
+      "epoch": 41.82,
+      "learning_rate": 1.4514141414141417e-06,
+      "loss": 1.036,
+      "step": 928500
+    },
+    {
+      "epoch": 41.84,
+      "learning_rate": 1.4413131313131315e-06,
+      "loss": 1.0354,
+      "step": 929000
+    },
+    {
+      "epoch": 41.87,
+      "learning_rate": 1.4312121212121214e-06,
+      "loss": 1.0349,
+      "step": 929500
+    },
+    {
+      "epoch": 41.89,
+      "learning_rate": 1.4211313131313132e-06,
+      "loss": 1.0351,
+      "step": 930000
+    },
+    {
+      "epoch": 41.91,
+      "learning_rate": 1.4110303030303033e-06,
+      "loss": 1.0353,
+      "step": 930500
+    },
+    {
+      "epoch": 41.93,
+      "learning_rate": 1.4009292929292931e-06,
+      "loss": 1.0373,
+      "step": 931000
+    },
+    {
+      "epoch": 41.96,
+      "learning_rate": 1.390828282828283e-06,
+      "loss": 1.0345,
+      "step": 931500
+    },
+    {
+      "epoch": 41.98,
+      "learning_rate": 1.3807272727272727e-06,
+      "loss": 1.0377,
+      "step": 932000
+    },
+    {
+      "epoch": 42.0,
+      "eval_accuracy": 0.7907370114861634,
+      "eval_loss": 0.8748875260353088,
+      "eval_runtime": 54.5624,
+      "eval_samples_per_second": 896.405,
+      "eval_steps_per_second": 14.021,
+      "step": 932484
+    },
+    {
+      "epoch": 42.0,
+      "learning_rate": 1.370626262626263e-06,
+      "loss": 1.0342,
+      "step": 932500
+    },
+    {
+      "epoch": 42.02,
+      "learning_rate": 1.3605252525252526e-06,
+      "loss": 1.0361,
+      "step": 933000
+    },
+    {
+      "epoch": 42.05,
+      "learning_rate": 1.3504242424242425e-06,
+      "loss": 1.0362,
+      "step": 933500
+    },
+    {
+      "epoch": 42.07,
+      "learning_rate": 1.3403434343434343e-06,
+      "loss": 1.0339,
+      "step": 934000
+    },
+    {
+      "epoch": 42.09,
+      "learning_rate": 1.3302424242424244e-06,
+      "loss": 1.0357,
+      "step": 934500
+    },
+    {
+      "epoch": 42.11,
+      "learning_rate": 1.3201414141414142e-06,
+      "loss": 1.0373,
+      "step": 935000
+    },
+    {
+      "epoch": 42.14,
+      "learning_rate": 1.310040404040404e-06,
+      "loss": 1.039,
+      "step": 935500
+    },
+    {
+      "epoch": 42.16,
+      "learning_rate": 1.299939393939394e-06,
+      "loss": 1.0341,
+      "step": 936000
+    },
+    {
+      "epoch": 42.18,
+      "learning_rate": 1.289838383838384e-06,
+      "loss": 1.0331,
+      "step": 936500
+    },
+    {
+      "epoch": 42.2,
+      "learning_rate": 1.2797373737373739e-06,
+      "loss": 1.0345,
+      "step": 937000
+    },
+    {
+      "epoch": 42.23,
+      "learning_rate": 1.2696363636363638e-06,
+      "loss": 1.0316,
+      "step": 937500
+    },
+    {
+      "epoch": 42.25,
+      "learning_rate": 1.2595555555555556e-06,
+      "loss": 1.0358,
+      "step": 938000
+    },
+    {
+      "epoch": 42.27,
+      "learning_rate": 1.2494545454545454e-06,
+      "loss": 1.0321,
+      "step": 938500
+    },
+    {
+      "epoch": 42.29,
+      "learning_rate": 1.2393535353535355e-06,
+      "loss": 1.0346,
+      "step": 939000
+    },
+    {
+      "epoch": 42.32,
+      "learning_rate": 1.2292525252525254e-06,
+      "loss": 1.036,
+      "step": 939500
+    },
+    {
+      "epoch": 42.34,
+      "learning_rate": 1.2191515151515152e-06,
+      "loss": 1.0362,
+      "step": 940000
+    },
+    {
+      "epoch": 42.36,
+      "learning_rate": 1.209050505050505e-06,
+      "loss": 1.0366,
+      "step": 940500
+    },
+    {
+      "epoch": 42.38,
+      "learning_rate": 1.1989494949494952e-06,
+      "loss": 1.0344,
+      "step": 941000
+    },
+    {
+      "epoch": 42.41,
+      "learning_rate": 1.1888484848484848e-06,
+      "loss": 1.0367,
+      "step": 941500
+    },
+    {
+      "epoch": 42.43,
+      "learning_rate": 1.178747474747475e-06,
+      "loss": 1.0307,
+      "step": 942000
+    },
+    {
+      "epoch": 42.45,
+      "learning_rate": 1.1686464646464648e-06,
+      "loss": 1.0364,
+      "step": 942500
+    },
+    {
+      "epoch": 42.47,
+      "learning_rate": 1.1585656565656566e-06,
+      "loss": 1.0344,
+      "step": 943000
+    },
+    {
+      "epoch": 42.5,
+      "learning_rate": 1.1484646464646465e-06,
+      "loss": 1.0351,
+      "step": 943500
+    },
+    {
+      "epoch": 42.52,
+      "learning_rate": 1.1383838383838385e-06,
+      "loss": 1.0341,
+      "step": 944000
+    },
+    {
+      "epoch": 42.54,
+      "learning_rate": 1.1282828282828284e-06,
+      "loss": 1.033,
+      "step": 944500
+    },
+    {
+      "epoch": 42.56,
+      "learning_rate": 1.1181818181818182e-06,
+      "loss": 1.0335,
+      "step": 945000
+    },
+    {
+      "epoch": 42.59,
+      "learning_rate": 1.1080808080808083e-06,
+      "loss": 1.0357,
+      "step": 945500
+    },
+    {
+      "epoch": 42.61,
+      "learning_rate": 1.0979797979797982e-06,
+      "loss": 1.0333,
+      "step": 946000
+    },
+    {
+      "epoch": 42.63,
+      "learning_rate": 1.087878787878788e-06,
+      "loss": 1.0327,
+      "step": 946500
+    },
+    {
+      "epoch": 42.65,
+      "learning_rate": 1.0777777777777779e-06,
+      "loss": 1.0362,
+      "step": 947000
+    },
+    {
+      "epoch": 42.68,
+      "learning_rate": 1.0676767676767677e-06,
+      "loss": 1.0353,
+      "step": 947500
+    },
+    {
+      "epoch": 42.7,
+      "learning_rate": 1.0575757575757576e-06,
+      "loss": 1.0319,
+      "step": 948000
+    },
+    {
+      "epoch": 42.72,
+      "learning_rate": 1.0474747474747475e-06,
+      "loss": 1.0353,
+      "step": 948500
+    },
+    {
+      "epoch": 42.74,
+      "learning_rate": 1.0373737373737375e-06,
+      "loss": 1.0351,
+      "step": 949000
+    },
+    {
+      "epoch": 42.77,
+      "learning_rate": 1.0272929292929294e-06,
+      "loss": 1.0339,
+      "step": 949500
+    },
+    {
+      "epoch": 42.79,
+      "learning_rate": 1.0171919191919192e-06,
+      "loss": 1.0357,
+      "step": 950000
+    },
+    {
+      "epoch": 42.81,
+      "learning_rate": 1.0070909090909093e-06,
+      "loss": 1.0361,
+      "step": 950500
+    },
+    {
+      "epoch": 42.83,
+      "learning_rate": 9.96989898989899e-07,
+      "loss": 1.0323,
+      "step": 951000
+    },
+    {
+      "epoch": 42.86,
+      "learning_rate": 9.86888888888889e-07,
+      "loss": 1.032,
+      "step": 951500
+    },
+    {
+      "epoch": 42.88,
+      "learning_rate": 9.768080808080809e-07,
+      "loss": 1.0339,
+      "step": 952000
+    },
+    {
+      "epoch": 42.9,
+      "learning_rate": 9.667070707070707e-07,
+      "loss": 1.0372,
+      "step": 952500
+    },
+    {
+      "epoch": 42.92,
+      "learning_rate": 9.566060606060606e-07,
+      "loss": 1.0354,
+      "step": 953000
+    },
+    {
+      "epoch": 42.95,
+      "learning_rate": 9.465050505050507e-07,
+      "loss": 1.0361,
+      "step": 953500
+    },
+    {
+      "epoch": 42.97,
+      "learning_rate": 9.364040404040404e-07,
+      "loss": 1.034,
+      "step": 954000
+    },
+    {
+      "epoch": 42.99,
+      "learning_rate": 9.263232323232325e-07,
+      "loss": 1.035,
+      "step": 954500
+    },
+    {
+      "epoch": 43.0,
+      "eval_accuracy": 0.7905800803288234,
+      "eval_loss": 0.8755584955215454,
+      "eval_runtime": 54.3889,
+      "eval_samples_per_second": 899.265,
+      "eval_steps_per_second": 14.065,
+      "step": 954686
+    },
+    {
+      "epoch": 43.01,
+      "learning_rate": 9.162222222222222e-07,
+      "loss": 1.0335,
+      "step": 955000
+    },
+    {
+      "epoch": 43.04,
+      "learning_rate": 9.061212121212122e-07,
+      "loss": 1.0328,
+      "step": 955500
+    },
+    {
+      "epoch": 43.06,
+      "learning_rate": 8.96020202020202e-07,
+      "loss": 1.0352,
+      "step": 956000
+    },
+    {
+      "epoch": 43.08,
+      "learning_rate": 8.85919191919192e-07,
+      "loss": 1.0348,
+      "step": 956500
+    },
+    {
+      "epoch": 43.1,
+      "learning_rate": 8.758181818181819e-07,
+      "loss": 1.0335,
+      "step": 957000
+    },
+    {
+      "epoch": 43.13,
+      "learning_rate": 8.657373737373738e-07,
+      "loss": 1.0337,
+      "step": 957500
+    },
+    {
+      "epoch": 43.15,
+      "learning_rate": 8.556363636363638e-07,
+      "loss": 1.0365,
+      "step": 958000
+    },
+    {
+      "epoch": 43.17,
+      "learning_rate": 8.455353535353536e-07,
+      "loss": 1.0332,
+      "step": 958500
+    },
+    {
+      "epoch": 43.19,
+      "learning_rate": 8.354545454545456e-07,
+      "loss": 1.0358,
+      "step": 959000
+    },
+    {
+      "epoch": 43.22,
+      "learning_rate": 8.253535353535353e-07,
+      "loss": 1.0337,
+      "step": 959500
+    },
+    {
+      "epoch": 43.24,
+      "learning_rate": 8.152525252525254e-07,
+      "loss": 1.0337,
+      "step": 960000
+    },
+    {
+      "epoch": 43.26,
+      "learning_rate": 8.051515151515152e-07,
+      "loss": 1.0349,
+      "step": 960500
+    },
+    {
+      "epoch": 43.28,
+      "learning_rate": 7.950505050505051e-07,
+      "loss": 1.0345,
+      "step": 961000
+    },
+    {
+      "epoch": 43.31,
+      "learning_rate": 7.84969696969697e-07,
+      "loss": 1.0331,
+      "step": 961500
+    },
+    {
+      "epoch": 43.33,
+      "learning_rate": 7.748686868686869e-07,
+      "loss": 1.0331,
+      "step": 962000
+    },
+    {
+      "epoch": 43.35,
+      "learning_rate": 7.647676767676768e-07,
+      "loss": 1.0349,
+      "step": 962500
+    },
+    {
+      "epoch": 43.37,
+      "learning_rate": 7.546666666666668e-07,
+      "loss": 1.0388,
+      "step": 963000
+    },
+    {
+      "epoch": 43.4,
+      "learning_rate": 7.445656565656566e-07,
+      "loss": 1.0352,
+      "step": 963500
+    },
+    {
+      "epoch": 43.42,
+      "learning_rate": 7.344646464646466e-07,
+      "loss": 1.0335,
+      "step": 964000
+    },
+    {
+      "epoch": 43.44,
+      "learning_rate": 7.243636363636363e-07,
+      "loss": 1.0349,
+      "step": 964500
+    },
+    {
+      "epoch": 43.46,
+      "learning_rate": 7.142626262626263e-07,
+      "loss": 1.0344,
+      "step": 965000
+    },
+    {
+      "epoch": 43.49,
+      "learning_rate": 7.041818181818183e-07,
+      "loss": 1.0335,
+      "step": 965500
+    },
+    {
+      "epoch": 43.51,
+      "learning_rate": 6.940808080808081e-07,
+      "loss": 1.0305,
+      "step": 966000
+    },
+    {
+      "epoch": 43.53,
+      "learning_rate": 6.839797979797981e-07,
+      "loss": 1.0348,
+      "step": 966500
+    },
+    {
+      "epoch": 43.55,
+      "learning_rate": 6.738787878787879e-07,
+      "loss": 1.0326,
+      "step": 967000
+    },
+    {
+      "epoch": 43.58,
+      "learning_rate": 6.637777777777778e-07,
+      "loss": 1.0347,
+      "step": 967500
+    },
+    {
+      "epoch": 43.6,
+      "learning_rate": 6.536767676767678e-07,
+      "loss": 1.0358,
+      "step": 968000
+    },
+    {
+      "epoch": 43.62,
+      "learning_rate": 6.435757575757575e-07,
+      "loss": 1.034,
+      "step": 968500
+    },
+    {
+      "epoch": 43.64,
+      "learning_rate": 6.334949494949496e-07,
+      "loss": 1.0331,
+      "step": 969000
+    },
+    {
+      "epoch": 43.67,
+      "learning_rate": 6.233939393939394e-07,
+      "loss": 1.032,
+      "step": 969500
+    },
+    {
+      "epoch": 43.69,
+      "learning_rate": 6.132929292929293e-07,
+      "loss": 1.035,
+      "step": 970000
+    },
+    {
+      "epoch": 43.71,
+      "learning_rate": 6.031919191919193e-07,
+      "loss": 1.034,
+      "step": 970500
+    },
+    {
+      "epoch": 43.73,
+      "learning_rate": 5.930909090909091e-07,
+      "loss": 1.0381,
+      "step": 971000
+    },
+    {
+      "epoch": 43.76,
+      "learning_rate": 5.82989898989899e-07,
+      "loss": 1.0355,
+      "step": 971500
+    },
+    {
+      "epoch": 43.78,
+      "learning_rate": 5.72888888888889e-07,
+      "loss": 1.0354,
+      "step": 972000
+    },
+    {
+      "epoch": 43.8,
+      "learning_rate": 5.627878787878788e-07,
+      "loss": 1.0359,
+      "step": 972500
+    },
+    {
+      "epoch": 43.82,
+      "learning_rate": 5.527070707070707e-07,
+      "loss": 1.035,
+      "step": 973000
+    },
+    {
+      "epoch": 43.85,
+      "learning_rate": 5.426060606060606e-07,
+      "loss": 1.0341,
+      "step": 973500
+    },
+    {
+      "epoch": 43.87,
+      "learning_rate": 5.325050505050505e-07,
+      "loss": 1.0343,
+      "step": 974000
+    },
+    {
+      "epoch": 43.89,
+      "learning_rate": 5.224040404040404e-07,
+      "loss": 1.0348,
+      "step": 974500
+    },
+    {
+      "epoch": 43.91,
+      "learning_rate": 5.123232323232324e-07,
+      "loss": 1.0353,
+      "step": 975000
+    },
+    {
+      "epoch": 43.94,
+      "learning_rate": 5.022222222222222e-07,
+      "loss": 1.0361,
+      "step": 975500
+    },
+    {
+      "epoch": 43.96,
+      "learning_rate": 4.921414141414142e-07,
+      "loss": 1.0336,
+      "step": 976000
+    },
+    {
+      "epoch": 43.98,
+      "learning_rate": 4.82040404040404e-07,
+      "loss": 1.0328,
+      "step": 976500
+    },
+    {
+      "epoch": 44.0,
+      "eval_accuracy": 0.7904323758427879,
+      "eval_loss": 0.8758755922317505,
+      "eval_runtime": 54.5925,
+      "eval_samples_per_second": 895.911,
+      "eval_steps_per_second": 14.013,
+      "step": 976888
+    },
+    {
+      "epoch": 44.01,
+      "learning_rate": 4.7193939393939395e-07,
+      "loss": 1.0363,
+      "step": 977000
+    },
+    {
+      "epoch": 44.03,
+      "learning_rate": 4.6183838383838386e-07,
+      "loss": 1.0334,
+      "step": 977500
+    },
+    {
+      "epoch": 44.05,
+      "learning_rate": 4.517373737373738e-07,
+      "loss": 1.0315,
+      "step": 978000
+    },
+    {
+      "epoch": 44.07,
+      "learning_rate": 4.4163636363636364e-07,
+      "loss": 1.037,
+      "step": 978500
+    },
+    {
+      "epoch": 44.1,
+      "learning_rate": 4.3153535353535356e-07,
+      "loss": 1.0335,
+      "step": 979000
+    },
+    {
+      "epoch": 44.12,
+      "learning_rate": 4.214343434343435e-07,
+      "loss": 1.0348,
+      "step": 979500
+    },
+    {
+      "epoch": 44.14,
+      "learning_rate": 4.113333333333334e-07,
+      "loss": 1.0342,
+      "step": 980000
+    },
+    {
+      "epoch": 44.16,
+      "learning_rate": 4.0123232323232325e-07,
+      "loss": 1.0337,
+      "step": 980500
+    },
+    {
+      "epoch": 44.19,
+      "learning_rate": 3.9113131313131317e-07,
+      "loss": 1.0342,
+      "step": 981000
+    },
+    {
+      "epoch": 44.21,
+      "learning_rate": 3.810303030303031e-07,
+      "loss": 1.0317,
+      "step": 981500
+    },
+    {
+      "epoch": 44.23,
+      "learning_rate": 3.7096969696969703e-07,
+      "loss": 1.0359,
+      "step": 982000
+    },
+    {
+      "epoch": 44.25,
+      "learning_rate": 3.608686868686869e-07,
+      "loss": 1.0321,
+      "step": 982500
+    },
+    {
+      "epoch": 44.28,
+      "learning_rate": 3.507676767676768e-07,
+      "loss": 1.0339,
+      "step": 983000
+    },
+    {
+      "epoch": 44.3,
+      "learning_rate": 3.406666666666667e-07,
+      "loss": 1.0339,
+      "step": 983500
+    },
+    {
+      "epoch": 44.32,
+      "learning_rate": 3.305858585858586e-07,
+      "loss": 1.0318,
+      "step": 984000
+    },
+    {
+      "epoch": 44.34,
+      "learning_rate": 3.204848484848485e-07,
+      "loss": 1.0327,
+      "step": 984500
+    },
+    {
+      "epoch": 44.37,
+      "learning_rate": 3.103838383838384e-07,
+      "loss": 1.032,
+      "step": 985000
+    },
+    {
+      "epoch": 44.39,
+      "learning_rate": 3.002828282828283e-07,
+      "loss": 1.0351,
+      "step": 985500
+    },
+    {
+      "epoch": 44.41,
+      "learning_rate": 2.901818181818182e-07,
+      "loss": 1.0372,
+      "step": 986000
+    },
+    {
+      "epoch": 44.43,
+      "learning_rate": 2.8008080808080813e-07,
+      "loss": 1.0328,
+      "step": 986500
+    },
+    {
+      "epoch": 44.46,
+      "learning_rate": 2.69979797979798e-07,
+      "loss": 1.0338,
+      "step": 987000
+    },
+    {
+      "epoch": 44.48,
+      "learning_rate": 2.598787878787879e-07,
+      "loss": 1.0323,
+      "step": 987500
+    },
+    {
+      "epoch": 44.5,
+      "learning_rate": 2.4979797979797984e-07,
+      "loss": 1.0349,
+      "step": 988000
+    },
+    {
+      "epoch": 44.52,
+      "learning_rate": 2.396969696969697e-07,
+      "loss": 1.033,
+      "step": 988500
+    },
+    {
+      "epoch": 44.55,
+      "learning_rate": 2.2959595959595962e-07,
+      "loss": 1.0342,
+      "step": 989000
+    },
+    {
+      "epoch": 44.57,
+      "learning_rate": 2.194949494949495e-07,
+      "loss": 1.0347,
+      "step": 989500
+    },
+    {
+      "epoch": 44.59,
+      "learning_rate": 2.093939393939394e-07,
+      "loss": 1.0326,
+      "step": 990000
+    },
+    {
+      "epoch": 44.61,
+      "learning_rate": 1.992929292929293e-07,
+      "loss": 1.0333,
+      "step": 990500
+    },
+    {
+      "epoch": 44.64,
+      "learning_rate": 1.891919191919192e-07,
+      "loss": 1.032,
+      "step": 991000
+    },
+    {
+      "epoch": 44.66,
+      "learning_rate": 1.790909090909091e-07,
+      "loss": 1.0348,
+      "step": 991500
+    },
+    {
+      "epoch": 44.68,
+      "learning_rate": 1.6901010101010103e-07,
+      "loss": 1.0357,
+      "step": 992000
+    },
+    {
+      "epoch": 44.7,
+      "learning_rate": 1.5890909090909091e-07,
+      "loss": 1.034,
+      "step": 992500
+    },
+    {
+      "epoch": 44.73,
+      "learning_rate": 1.4880808080808083e-07,
+      "loss": 1.0352,
+      "step": 993000
+    },
+    {
+      "epoch": 44.75,
+      "learning_rate": 1.3870707070707072e-07,
+      "loss": 1.0348,
+      "step": 993500
+    },
+    {
+      "epoch": 44.77,
+      "learning_rate": 1.286060606060606e-07,
+      "loss": 1.0309,
+      "step": 994000
+    },
+    {
+      "epoch": 44.79,
+      "learning_rate": 1.1854545454545456e-07,
+      "loss": 1.0331,
+      "step": 994500
+    },
+    {
+      "epoch": 44.82,
+      "learning_rate": 1.0844444444444446e-07,
+      "loss": 1.0346,
+      "step": 995000
+    },
+    {
+      "epoch": 44.84,
+      "learning_rate": 9.834343434343436e-08,
+      "loss": 1.0334,
+      "step": 995500
+    },
+    {
+      "epoch": 44.86,
+      "learning_rate": 8.824242424242425e-08,
+      "loss": 1.0334,
+      "step": 996000
+    },
+    {
+      "epoch": 44.88,
+      "learning_rate": 7.814141414141414e-08,
+      "loss": 1.0311,
+      "step": 996500
+    },
+    {
+      "epoch": 44.91,
+      "learning_rate": 6.804040404040404e-08,
+      "loss": 1.032,
+      "step": 997000
+    },
+    {
+      "epoch": 44.93,
+      "learning_rate": 5.795959595959596e-08,
+      "loss": 1.0353,
+      "step": 997500
+    },
+    {
+      "epoch": 44.95,
+      "learning_rate": 4.785858585858586e-08,
+      "loss": 1.0344,
+      "step": 998000
+    },
+    {
+      "epoch": 44.97,
+      "learning_rate": 3.775757575757576e-08,
+      "loss": 1.0332,
+      "step": 998500
+    },
+    {
+      "epoch": 45.0,
+      "learning_rate": 2.7656565656565656e-08,
+      "loss": 1.0367,
+      "step": 999000
+    },
+    {
+      "epoch": 45.0,
+      "eval_accuracy": 0.7904324060897042,
+      "eval_loss": 0.8771223425865173,
+      "eval_runtime": 53.6883,
+      "eval_samples_per_second": 910.999,
+      "eval_steps_per_second": 14.249,
+      "step": 999090
+    },
+    {
+      "epoch": 45.02,
+      "learning_rate": 1.755555555555556e-08,
+      "loss": 1.0352,
+      "step": 999500
+    },
+    {
+      "epoch": 45.04,
+      "learning_rate": 7.494949494949496e-09,
+      "loss": 1.0317,
+      "step": 1000000
+    },
+    {
+      "epoch": 45.04,
+      "eval_accuracy": 0.7908125411117414,
+      "eval_loss": 0.8743389248847961,
+      "eval_runtime": 52.6097,
+      "eval_samples_per_second": 929.676,
+      "eval_steps_per_second": 14.541,
+      "step": 1000000
+    },
+    {
+      "epoch": 45.04,
+      "step": 1000000,
+      "total_flos": 6.737894813932611e+19,
+      "train_loss": 0.020673565490722655,
+      "train_runtime": 11326.2724,
+      "train_samples_per_second": 22602.317,
+      "train_steps_per_second": 88.29
+    }
+  ],
+  "max_steps": 1000000,
+  "num_train_epochs": 46,
+  "total_flos": 6.737894813932611e+19,
+  "trial_name": null,
+  "trial_params": null
+}